English – FPT AI

1. Introduction

 FPT Smart Cloud Company Limited (“FPT FPT Smart Cloud” hereinafter) Personal Data Protection Policy, privacy statement, procedures, guidelines, and templates lay out strict requirements for processing personal data pertaining to customers, business partners, employees or any other individual. It meets the requirements of the European Data Protection Regulation, Personal Data Protection Decree No. 13/2023/ND-CP as well as other national Data Protection Regulations and ensures compliance with the principles of national and international data protection laws in force all over the world. The policy, privacy statement, procedures, guidelines, and templates set a globally applicable data protection and security standard for FPT Smart Cloud and regulates the sharing of information between FPT Smart Cloud, subsidiaries, legal entities, and partners. FPT Smart Cloud has established guiding data protection principles – among them transparency, data economy and data security – as FPT Smart Cloud guidelines.  

1.1. Purpose

 The FPT Smart Cloud Personal Data Protection Policy, and privacy statement applies worldwide to FPT Smart Cloud, subsidiaries as well legal entities and is based on globally accepted, basic principles on data protection. Ensuring data protection is the foundation of trustworthy business relationships and the reputation of FPT Smart Cloud as a first-class employer.

The Personal Data Protection Policy provides one of the necessary framework conditions for cross-border data transfer among FPT Smart Cloud, Subsidiaries, and legal entities. It ensures the adequate level of data protection prescribed by the European Union General Data Protection Regulation, Protection of Personal Data Decree No. 13/2023/ND-CP or other national Personal Data Protection Regulations and the national laws for cross-border data transmission, including in countries that do not yet have adequate data protection laws. 

To standardize the collection, processing, transfer, and use of personal data, and promote the reasonable, lawfully, fairly, and transparent use of personal data to prevent personal data from being stolen, altered, damaged, lost or leaked, FPT Smart Cloud establishes the Personal Data Protection Policy, Privacy Statement, and information security policies.

1.2. Application Scope

All processing of personal data by FPT Smart Cloud is within the scope of this procedure.

This policy is binding for all departments and functions globally which are involved in personal identifiable information processing. Every FPT Smart Cloud department, legal entity or subsidiary must follow this procedure.

In scope are all data subjects whose personal data is collected, in line with the requirements of the Protection of Personal Data Decree No. 13/2023/ND-CP, GDPR and other national/ international data protection regulation.

1.3. Application of national Laws

The Personal Data Protection Policy, privacy statement, procedures, guidelines, and templates comprise the internationally accepted data privacy principles without replacing the existing national/international laws. It supplements the national data privacy laws. The relevant national law will take precedence in the event that it conflicts with the Personal Data Protection Policy and guidelines, or it has stricter requirements than this Policy and guidelines. The content of the Personal Data Protection Policy, procedures and guidelines must also be observed in the absence of corresponding national legislation. The reporting requirements for data processing under national laws must be observed. 

Each subsidiary or legal entity of FPT Smart Cloud is responsible for compliance with the Personal Data Protection Policy, this privacy statement, guidelines, and the legal obligations. If there is reason to believe that legal obligations contradict the duties under the Personal Data Protection Policy, privacy statement, procedures or the guidelines, the relevant subsidiary or legal entity must inform the Data Protection Officer. In the event of conflicts between national legislation, the Personal Data Protection Policy, and this privacy statement, FPT Smart Cloud will work with the relevant subsidiary or legal entity of FPT Smart Cloud to find a practical solution that meets the purpose of the Personal Data Protection Policy, guidelines, and this procedure. 

 1.4. Responsibilities

The Data Protection Officer is responsible for ensuring that the privacy statement is correct and that mechanisms exist such as having the privacy statement on FPT Smart Cloud website to make all data subjects aware of the contents of this notice prior FPT Smart Cloud commencing collection of their data. 

The Data Protection Officer is responsible for ensuring that this statement is made available to data subjects prior to FPT Smart Cloud collecting/processing their personal data.

All Employees/ Staff of FPT Smart Cloud who interact with data subjects are responsible for ensuring that this statement is drawn to the data subject’s attention and their consent to the processing of their data is secured.

2. Privacy Statement

FPT Smart Cloud is part of FPT Corporation (FPT – HoSE) – the global leading technology and IT services group headquartered in Vietnam with nearly US$2.5 billion in revenue and 54,687 employees. Qualified with ISO 9001: 2015, ISO 27001:2022, ISO 27027: 2015; ISO 27018: 2019, PCI DSS, FPT Smart Cloud delivers world-class services in Cloud Computing services, Artificial Intelligence (Al) services, AI Infrastructure, AI Platform, AI as a Service, Data as a Service and Consolidation of Financial Statements solution globally from delivery centers across the Japan, Vietnam and the Asia Pacific.

 Personal data type

Name, email address, designation, company, country and telephone number

IP address, demographics, your device operating system, and browser type

Source (FPT Smart Cloud obtained the personal data from if it has not been collected directly from you, the data subject)

FPT Smart Cloud WEB page

2.1. Personal Information we may collect and process

You can assess or visit our website at any time without informing us who you are or providing us any personal information. However, we may collect information at our websites in two ways: (1) directly (for example, when you provide information, such as your name, email address, designation, company, country and telephone number, to sign up for a newsletter or register to comment on a forum website); and (2) indirectly (for example, through our website’s technology, we may collect certain information such as your IP address, demographics, your computers’ operating system, and browser type).

We do not attempt to track your personal information in order to identify you, but gathering these contact information in order to make up the web traffic routing, to diagnose problems with server for administration of our website, to better understand how you interact with our website and services and to re-design and upgrade the website for better use. If you choose not to provide your personal information that is mandatory to process your request, we may not be able to provide the corresponding service.

  2.2. Use of collected information

We use personal data to provide you with information you request, process online job applications, and for other purposes which we would describe to you at the point where it is collected or which will be obvious to you. For example:

To further fulfil your requirements on products and services

To contact you with the aim of developing a business relationship

To feedback to your idea and/or to provide you relevant information at your requirements

To contact you for marketing purpose such as customer surveys

To inform you about our company

To obey regulations in applicable laws

2.3. Consent

By consenting to this privacy notice you are giving us permission to process your personal data specifically for the purposes identified.

Consent is required for FPT Smart Cloud to process personal data, but it must be explicitly given. Where we ask you for personal data, we will always tell you why and how the information will be used. 

Means: FPT Smart Cloud will inform you about the purpose of the processing, contact details of the Data controller or its representative, lawful basis of the processing, personal data was obtained, if not obtained directly from the data subject.

FPT Smart Cloud provides updated information without any undue delay and before continuing with the processing if the purposes for the processing of the personal data are changed or extended. In this case FPT Smart Cloud will ask for a new consent.

You may withdraw consent at any time by email, a written letter or telephone call to our Data Protection Officer.

2.4. Data recipients, transfer, and disclosure of personal information

We do not share your personal information with third parties without seeking your prior permission. We will seek your consent prior to using or sharing personal information for any purpose beyond the requirement for which it was originally collected. However, we may share your personal information within FPT Smart Cloud or with any of its subsidiaries, business partners, service vendors, authorized third-party agents, or contractors located in any part of the world for the purposes of data processing, storage, or to provide a requested service or transaction, after ensuring that such entities are contractually bound by data privacy obligations. When required, we may disclose personal information to external law enforcement bodies or regulatory authorities, in order to comply with legal obligations.

We do not intend for our websites or online services to be used by anyone under the age of 13. If you are a parent or guardian and believe we may have collected information about a child, please contact us as described in this Privacy Statement.

FPT Smart Cloud considers that, as a general rule, a child of 16 and over is mature enough to understand giving of consent, they are giving and should be in a position to give that consent. All Data subjects will be required to verify their identity. Where personal data is sought in respect of a child below the age of 16, a parent or guardian must give the consent on behalf of the child. Any response will be directed to the parent or guardian. FPT Smart Cloud will need to be satisfied as to the identity of the parent or guardian, and that they are acting in the best interests of the child, before excepting the consent in respect of the child. Parent or guardian has the obligation to explain the process and the content to the child and if it is legally required (PERSONAL DATA PROTECTION DECREE NO. 13/2023/ND-CP) to get the consent of a child, it is parent, agent or guardian responsibility.

If parent applying on behalf of a child under 16 years of age, FPT Smart Cloud will require proof of identity and address of parent and that of the child, together with the birth certificate of the child.
If a legal guardian applying on behalf of a child under 16 years of age, FPT Smart Cloud will require proof of guardian identity and address and that of the data subject, together with proof of authority to act as legal guardian and the birth certificate of the child.

If you are an agent acting on someone’s behalf (e.g. a solicitor applying on behalf of a client), FPT Smart Cloud may require proof of agent identity and address and that of the data subject, and proof that the data subject has given consent to act on their behalf.

 2.5. Disclosure

FPT Smart Cloud will pass on your personal data to third parties.

Third country (non-EU) / international organisation:

FPT Smart Cloud subsidiaries and legal entities globally

Safeguards in place to protect your personal data:

Processing agreement including Standard Contract Clause

Retrieve a copy of the safeguards in place here:

Data Protection Officer 

2.6. Retention period

FPT Smart Cloud will process personal data for one year. Retention period 2 years or based on applicable national laws/regulations.

2.7. Cookies policy

Like many websites, when you access to our websites, we will use “website assessment diary”- a cookie technology to collect additional website usage data. A cookie is a small data file that we transfer to your computer to facilitate your assessment to our websites. We may use information collected from our cookies to identify user behavior and to serve content and offers based on your profile, and for the other purposes described below, to the extent legally permissible in certain jurisdictions. In addition, when you visit our websites, our advertisement partners, whom we have engaged for re-marketing, may introduce cookies. Based on your browsing of our website you may see our advertisements while browsing through our advertisement partner websites and/or their network websites.

Such cookies would allow us to monitor the effectiveness of the advertisements and to make the advertisements more relevant to you. By using our site, you agree that we can place cookies on your device as explained herein. If you want to remove existing cookies from your device, you can do this using your browser options. Most Internet browsers automatically accept cookies. You can instruct your browser, by editing its options, to stop accepting cookies or to prompt you before accepting a cookie from the websites you visit.

2.8. Data Security

FPT Smart Cloud commits to secure your personal information with securities measures in place. The measures will help protecting data from the misuse, loss, leakage and/or alteration of information. Your personal information is access restricted to authorize FPT Smart Cloud’s personnel for the sake of providing service at your requirements and/or for FPT Smart Cloud’s audit, internal audit and for the purpose of law obligation. We strictly require our personnel, in any way, to protect your personal information and have use all measurements, technology and recognized security process for this purpose in compliance with government authorizations’ regulations. Regarding your use of our websites, you should understand that the open nature of the Internet is such that information and personal data may flow over networks connecting you to our systems without security measures and may be accessed and used by people other than those for whom the data is intended.

2.9. Links to other websites

This site contains links to other websites, but they are neither FPT Smart Cloud’s websites nor under control of FPT Smart Cloud. FPT Smart Cloud is not responsible for the privacy practices or the content and transactions of such websites. You are required to read carefully the Privacy part of those linked websites to assure that you have fully understood the way of personal information collection and sharing before providing your own information. You shall take all responsibility of risk that may incur when using those websites.

2.10. Your rights as a data subject

At any point while we are in possession of or processing your personal data, you, the data subject, have the following rights:

Right to be informed – you have the right to request information what kind of your personal data are collect, use, processed, for what purpose, from which source, lawful basis of processing

Right of access – you have the right to request a copy of the information that we hold about you.

Right of rectification – you have a right to correct data that we hold about you that is inaccurate or incomplete.

Right to be forgotten/erasure – in certain circumstances you can ask for the data we hold about you to be erased from our records.

Right to restriction of processing – where certain conditions apply to have a right to restrict the processing.

Right of portability – you have the right to have the data we hold about you transferred to another organisation.

Right to object – you have the right to object to certain types of processing such as direct marketing.

Right to object to automated processing, including profiling – you also have the right to be subject to the legal effects of automated processing or profiling.

Right to judicial review: if FPT Smart Cloud refuses your request under rights of access, we will provide you with a reason as to why. You have the right to complain as outlined in below.

Right to claim damages: The data subject has the right to claim damage as prescribed by law when there are violations against regulations on protection of his/her personal data, unless otherwise agreed by parties or unless otherwise prescribed by law.

Right to self-protection: The data subject has the right to self-protection according to regulations in the Civil Code, other relevant laws and this Decree, or request competent agencies and organizations to implement civil right protection methods according to regulations in Article 11 of the Civil Code.

All the above requests will be forwarded on should there be a third party involved in the processing of your personal data.

FPT Smart Cloud accepts the following forms of ID when information on your personal data or data subject rights is requested: Passport, driving license, ID card.

2.11. Complaints

If you wish to make a complaint about how your personal data is being processed by FPT Smart Cloud or how your complaint has been handled, you have the right to lodge a complaint directly with the supervisory authority and FPT Smart Cloud’s Data Protection Officer.

2.12. Contact details

Supervisory authority Vietnam contact details:

Contact name: Ministry of public security.

Address: 30 Tran Binh Trong, Hai Ba Trung Ward, Ha Noi, Vietnam

Telephone: + 84 692343647

Data Protection Officer (DPO):

Contact name: Pham The Minh.

Address: FPT Tower, 10 Pham Van Bach Street, Cau Giay Ward, Ha Noi, Vietnam

Email: MinhPT@fpt.com

Telephone: +84 913571357

Contact details of other countries supervisory authorities you can get form DPO at any time without any undue delay.

2.13. Changes on Privacy Statements

FPT Smart Cloud reserves the rights to change, modify, add or remove in whole or in part this Privacy Statement at its sole discretion, at any time. Therefore, you are responsible for regularly reviewing this statement. Changes of this Privacy Statements will be posted on this website. These changes will also be effective when they are posted. Your continued use of this statement constitutes your agreement to all such terms.

2.14. Contact

If you have any questions about our Privacy Statement or about how to protect your personal information, you can contact the Data Protection Officer of FPT Smart Cloud. 
Data Protection Officer: Mr. Pham The Minh, Data Protection Officer.

Address: FPT Tower, 10 Pham Van Bach Street, Cau Giay Ward, Ha Noi, Vietnam.

Email: MinhPT@fpt.com.

Telephone: +84 913571357.

2.15. Document Owner and Approval

The Data Protection Officer (DPO) is the owner of this document and is responsible for ensuring that this statement is reviewed in line with the review requirements of the Personal Data Protection Policy.

This statement was approved by a Board member responsible for Data Protection. 

3. Appendix

 3.1. Definition

Abbreviations

Description

PII, Personal Identifiable Information, Personal Data

Refer to the personal data defined by the EU GDPR (Article 4 (1)), ‘personal data’ means any information relating to an identified  or identifiable natural person (‘data  subject’); an identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person 

Data Subject

EU GDPR (Article 4 – 1),
Data subject refers to any individual person who can be identified, directly or indirectly.

Data Controller

EU GDPR (Article 4 – 7),
Refer to the personal data defined by the EU GDPR (Article 4 (1)), ‘personal data’ means any information relating to an identified  or identifiable natural person (‘data  subject’); an identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person.

Data Processor

EU GDPR (Article 4 – 8),
Data Processor means a natural or legal person, public authority, agency or anybody which processes data on behalf of the controller.

Recipient

EU GDPR (Article 4 – 9),
A natural or legal person, public authority, agency or anybody, to which the personal data are disclosed, whether third party or not.

Third Party

EU GDPR (Article 4 – 10),
A natural or legal person, public authority, agency or anybody other than the data subject, controller, processor and persons who under direct authority of controller or processor, are authorized to process personal data

DPO

Data Protection Officer

DPIA

Data Protection Impacted Assessment

European Union

3.2. Related Documents

Code

Name of documents

EU GDPR

EU General Data Protection Regulation

PERSONAL DATA PROTECTION DECREE NO. 13/2023/ND-CP, VN

Decree of the Vietnamese Government: PERSONAL DATA PROTECTION DECREE NO. 13/2023/ND-CP
Nghị Định Quy Định Về Bảo Vệ Dữ Liệu Cá Nhân 07/2023

PCI DSS

Payment Card Industry Data Security Standard,

3.3. Data Protection Law, Vietnam, Overview

There is no single data protection law in Vietnam. Regulations on data protection and privacy can be found in various legal instruments. The right of privacy and right of reputation, dignity and honour and fundamental principles of such rights are currently provided for in Constitution 2013 (“Constitution”) and Civil Code 2015 (“Civil Code”) as inviolable and protected by law.
Regarding personal data, the guiding principles on collection, storage, use, process, disclosure or transfer of personal information are specified in the following main laws and documents:

Data Law No. 60/2024/QH15, passed by the National Assembly on 30 November 2024. This Law comes into force as of July 1, 2025.

Criminal Code No. 100/2015/QH13, passed by the National Assembly on 27 November 2015

Law No. 24/2018/QH14 on Cybersecurity, passed by the National Assembly on 12 June 2018 (“Cybersecurity Law”);

Law No. 86/2015/QH13 on Network Information Security, passed by the National Assembly on 19 November 2015; as amended by Law No. 35/2018/QH14 dated 20 November 2018, on amendments to some articles concerning planning of 37 Laws (“Network Information Security Law”);

Law No. 59/2010/QH12 on Protection of Consumers’ Rights, passed by the National Assembly on 17 November 2010; as amended by Law No.35/2018/QH14 dated 20 November 2018, on amendments to some articles concerning planning of 37 Laws (“CRPL”);

Law No. 67/2006/QH11 on Information Technology, passed by the National Assembly on 29 June 2006; as amended by Law No. 21/2017/QH14 dated 14 November 2017 on planning (“IT Law”);

Law No. 51/2005/QH11 on E-transactions, passed by the National Assembly on 29 November 2005 (“E-transactions Law”);

Decree No. 85/2016/ND-CP dated 1 July 2016, on the security of information systems by classification (“Decree 85”);

Decree No. 72/2013/ND-CP dated 15 July 2013 of the Government, on management, provision and use of Internet services and online information; as amended by Decree No. 27/2018/ND-CP dated 1 March 2018 and Decree No.150/2018/ND-CP dated 7 November 2018 (“Decree 72”);

Decree No. 52/2013/ND-CP dated 16 May 2013 of the Government; as amended by Decree No. 08/2018/ND-CP dated 15 January 2018, on amendments to certain Decrees related to business conditions under state management of the Ministry of Industry and Trade and Decree No. 85/2021/ND-CP dated 25 September 2021 (“Decree 52”);

Decree No. 15/2020/ND-CP of the Government dated 3 February 2020 on penalties for administrative violations against regulations on postal services, telecommunications, radio frequencies, information technology and electronic transactions (“Decree 15”);

Circular No. 03/2017/TT-BTTTT of the Ministry of Information and Communications dated 24 April 2017 on guidelines for Decree 85 (“Circular 03”);

Circular No. 20/2017/TT-BTTTT dated 12 September 2017 of the Ministry of Information and Communications, providing for Regulations on coordinating and responding to information security incidents nationwide (“Circular 20”);

Circular No. 38/2016/TT-BTTTT dated 26 December 2016 of the Ministry of Information and Communications, detailing cross-border provision of public information (“Circular 38”);

Circular No. 24/2015/TT-BTTTT dated 18 August 2015 of the Ministry of Information and Communications, providing for the management and use of Internet resources, as amended by Circular No. 06/2019/TT-BTTTT dated 19 July 2019 (“Circular 25”); and

Decision No. 05/2017/QD-TTg of the Prime Minister dated 16 March 2017 on emergency response plans to ensure national cyber-information security (“Decision 05”).

Applicability of the legal documents will depend on the factual context of each case, e.g businesses in the banking and finance, education, healthcare sectors may be subject to specialized data protection regulations, not to mention to regulations on employees’ personal information as provided in Labour Code 2019 (“Labour Code”).

Any risk and violation you want to inform, please report immediately by following:

Data Protection Officer, Pham The Minh

FPT Tower, 10 Pham Van Bach Street, Cau Giay Ward, Ha Noi, Viet Nam 

Cell: +84 913571357

E-mail: MinhPT@fpt.com

1. Introduction

FPT Smart Cloud Limited Company (“FPT Smart Cloud” hereinafter) Corporate Data Protection Policy lays out strict requirements for processing personal data pertaining to customers, business partners, employees or any other individual. It meets the requirements of the European Data Protection Directive and ensures compliance with the principles of national and international data protection laws in force all over the world. The policy sets a globally applicable data protection and security standard for FPT Smart Cloud and regulates the sharing of information between FPT Smart Cloud, subsidiaries, and legal entities. FPT Smart Cloud have established guiding data protection principles – among them transparency, data economy and data security – as FPT Smart Cloud Personal Data Protection Policy and Information security management guidelines. 

FPT Smart Cloud managers and employees are obligated to adhere to the Corporate Data Protection Policy and observe their local data protection laws. As the Data Protection Officer, it is my duty to ensure that the rules and principles of data protection at FPT Smart Cloud are followed around the world. 

I will be pleased to answer any questions you have about data protection and international personal data transfer.

Pham The Minh

Data Protection Officer, MinhPT@fpt.com, +84 913571357

1.1. Purpose

 This Data Protection Policy applies worldwide to FPT Smart Cloud, Subsidiaries as well as legal entities and is based on globally accepted, basic principles of data protection. Ensuring data protection is the foundation of trustworthy business relationships and the reputation of the FPT Smart Cloud as a first-class employer. 
The Data Protection Policy provides one of the necessary framework conditions for cross-border data transfer among FPT Smart Cloud, subsidiaries, and legal entities. It ensures an adequate level of data protection prescribed by the European Union General Data Protection Regulation, Personal Data Protection Decree No. 13, APPI, PDPA or other national Personal Data Protection Regulations and national laws for cross-border data transmission, including to countries which do not have adequate data protection law, yet. 
In order to standardize the collection, processing, transfer, and use of personal data, and promote the reasonable, lawfully, fairly and transparent use of personal data to prevent personal data from being stolen, altered, damaged, lost or leaked, FPT Smart Cloud establishes the personal data protection management policy and information security policies.

 1.2. Application Scope

 All processing of personal data by FPT Smart Cloud is within the scope of this procedure.

Means, all FPT Smart Cloud’s business processes and information systems involved in the collection, processing, use and transfer of personal data and all employees, contractors and 3rd party providers involved in the processing of personal data on behalf of FPT Smart Cloud.
This policy is binding for all departments and functions globally which are involved in personal identifiable information processing. Every FPT Smart Cloud department, legal entity or subsidiary must follow this procedure.
In scope are all data subjects whose personal data is collected, in line with the requirements of the GDPR, Personal Data Protection Decree No. 13 and other national/ international data protection regulation. 

1.3. Application of national Laws

This Data Protection Policy comprises the internationally accepted data privacy principles without replacing the existing national laws. It supplements the national data privacy laws. The relevant national law will take precedence in the event that it conflicts with this Data Protection Policy, or it has stricter requirements than this Policy. The content of this Data Protection Policy must also be observed in the absence of corresponding national legislation. The reporting requirements for data processing under national laws must be observed. 
Each subsidiary or legal entity of FPT Smart Cloud is responsible for compliance with this Data Protection Policy and the legal obligations. If there is reason to believe that legal obligations contradict the duties under this Data Protection Policy, the relevant subsidiary or legal entity must inform the Data Protection Officer. In the event of conflicts between national legislation and the Data Protection Policy, FPT Smart Cloud in person the Data Protection Officer will work with the relevant subsidiary or legal entity of FPT Smart Cloud to find a practical solution that meets the purpose of the Data Protection Policy.

1.4 Prevention of national and international Data Protection Laws Violations

 The Data Protection Officer DPO reporting to the board member responsible for Data Protection oversees the compliance and regulatory functions FPT Smart Cloud, with the goal to identify, reduce, and monitor all areas of possible regulatory and reputational risk regarding personal data processing.
The Personal Data Protection Policy and guidelines, procedures, templates is revised and supplemented once a year. The DPO and board member reviews and approves the Handbook promptly in the event of any material change in laws. regulations or business practices.
DPO provides periodically an online personal data protection education programs on online training platform to keep employees informed about current regulatory developments, updates of policies and procedures, and legal requirements.
If a violation of the Personal Data Protection policies, guidelines, procedures, templates occurs or a preliminary determination is made that a violation may have occurred, a report must be made to the DPO and Senior Management.
The Senior Management should impose adequate sanctions on employees violating the policies. Sanctions may include any or all of the following: a letter of censure, a fine, temporary suspension of employment, termination of employment, or any other sanction deemed appropriate by Senior Management.

2. Policy

 2.1. Guiding principles

2.2.1 Rules for protection of personal data

The personal data shall be processed as prescribed by law.
Data subjects are informed about activities related to the processing of their personal data, unless otherwise provided for by law.
The personal data shall be processed for the purposes that have been registered and declared by the Personal Data Controller, the Personal Data Processor, the Personal Data Controller-cum-Processor and the Third Party.
The collected personal data shall be appropriate for the scope and purposes of processing.  The purchase or sale of personal data shall be prohibited in any form, unless otherwise provided for by law.
The personal data shall be updated and added for the processing purposes.
The personal data shall be protected and secured throughout the processing. To be specific, the personal data shall be protected from violations against regulations on protection of personal data and prevention of loss, destruction or damage caused by incidents and use of technical measures.
The personal data shall be stored within a period of time that is appropriate for the processing purposes, unless otherwise provided for by law.
The Personal Data Controller and the Personal Data Controller-cum-Processor shall comply with the rules for data processing specified in Clauses 1 through 7 of this Article and prove their compliance.

2.2.2 Ensuring Data Subject’s Rights

Right to be informed

The data subject has the right to be informed of his/her personal data processing, unless otherwise provided for by law.

Right to give consent

The data subject has the right to give consent to the processing of his/her personal data, other than cases specified in Article 17 of Decree No 13/2023/NĐ-CP.

Right to access personal data

The data subject has the right to access his/her personal data in order to look at, rectify or request rectification of his/her personal data, unless otherwise provided for by law.

Right to withdraw consent

The data subject has the right to withdraw his/her consent, unless otherwise provided for by law.

Right to delete personal data

The data subject has the right to delete or request deletion of his/her personal data, unless otherwise provided for by law.

Right to obtain restriction on processing
The Data Subject has the right to obtain restriction on the processing of his/her personal data, unless otherwise provided for by law.
The restriction of data processing shall be carried out within 72 hours after the request of the Data Subject, with respect to all personal data requested by the data subject, unless otherwise provided for by law.
Right to obtain personal data

The Data Subject has the right to request the Personal Data Controller and the Personal Data Controller-cum-Processor to provide him/her with his/her personal data, unless otherwise provided for by law.

Right to object to processing

a) The data subject has the right to object to the Personal Data Controller and the Personal Data Controller-cum-Processor processing his/her personal data in order to prevent or restrict the ddisclosure of personal data or the use of personal data for advertising and marketing purposes, unless otherwise provided for by law.
b) The Personal Data Controller and the Personal Data Controller-cum-Processor shall comply with the data subject’s request within 72 hours after receiving the request, unless otherwise provided for by law.
Right to file complaints, denunciations, and lawsuits

The Data Subject has the right to file complaints, denunciations and lawsuits as prescribed by law.

Right to claim damage

The Data Subject has the right to claim damage as prescribed by law when there are violations against regulations on protection of his/her personal data, unless otherwise agreed by parties or unless otherwise prescribed by law.

The right to self-protection

The Data Subject has the right to self-protection according to regulations in the Civil Code, other relevant laws and Decree No 13/2023/NĐ-CP, or request competent agencies and organizations to implement civil right protection methods according to regulations in Article 11 of the Civil Code.

2.2. Customer and Provider Data (3rd party)

2.2.1 Data processing for a contractual relationship

Personal Data of customers and providers (3rd party) can be processed in order to establish, execute and terminate a contract. Prior to a contract – during the contract initiation phase – Personal Data can be processed to prepare bids or purchase orders or to fulfill other requests that relate to contract conclusion. Customers or providers can be contacted during the contract preparation process using the information that they have provided. Any restrictions requested by customers or providers must be complied with.

FPT Smart Cloud does not need the consent of the Data Subject to perform contractual obligations. 

The public, means every customer, provider, data subjects must have access to information about the FPT Smart Cloud’s Personal Data Protection principles and activities and must be able to communicate with FPT Smart Cloud’s Data Protection Officer in an easy way: 

Pham The Minh | Data Protection Officer, | FPT SMART CLOUD

Address: FPT Tower, 10 Pham Van Bach Street, Cau Giay Ward, Ha Noi, Vietnam
Cell: +84 913571357 | Tel: 1900638399
URL: https://fptsmartcloud.com/

2.2.2 Consent to data processing

Data can be processed following consent by the Data Subject. Before giving consent, the data subject must be informed in accordance with company’s Personal Data Protection Policy. In order to obtain the consent of the data subject, the following contents must be notified to the data subject:

a) The type of personal data to be processed;

b) Purpose of processing personal data;

c) Organizations and individuals may process personal data;

d) Rights and obligations of the data subject.

The declaration of consent must be obtained in writing or electronically for the purposes of documentation. In some circumstances, such as telephone conversations, consent can be given verbally. The granting of consent must be documented.

2.2.3 Data processing pursuant to legal authorization

The processing of personal data is also permitted if national legislation requests, requires or allows this. The type and extent of data processing must be necessary for the legally authorized data processing activity and must comply with the relevant statutory provisions.  

2.2.4 Data processing pursuant to legitimate interest

Personal Data can also be processed if it is necessary for a legitimate interest of FPT Smart Cloud. Legitimate interests are generally of a legal (e.g. collection of outstanding receivables) or commercial nature (e.g. avoiding breaches of contract). Personal Data may not be processed for the purposes of a legitimate interest if, in individual cases, there is evidence that the interests of the data subject merit protection, and that this takes precedence. Before data is processed, it is necessary to determine whether there are interests that merit protection.

2.2.5 User data and internet

If Personal Data is collected, processed and used on websites or in apps, the data subjects must be informed of this in a privacy statement and, if applicable, information about cookies. The privacy statement and any cookie information must be integrated so that it is easy to identify, directly accessible and consistently available for the data subjects.

If use profiles (tracking) are created to evaluate the use of websites and apps, the data subjects must always be informed accordingly in the privacy statement.

If websites or apps can access Personal Data in an area restricted to registered users, the identification and authentication of the data subject must offer sufficient protection during access. 

2.3. Employee Data

2.3.1 Data processing for the employment relationship

In employment relationships, personal data can be processed if needed to initiate, carry out and terminate the employment agreement. When initiating an employment relationship, the applicants’ personal data can be processed. If the candidate is rejected, his/her data must be deleted in observance of the required retention period, unless the applicant has agreed to remain on file for a future selection process. Consent must be given by every candidate before processing their personal data in FPT Smart Cloud systems. Consent is also needed to use the data for further application processes or before sharing the application with other FPT Smart Cloud legal entities. 

In the existing employment relationship, data processing must always relate to the purpose of the employment agreement if none of the following circumstances for authorized data processing apply.

If it should be necessary during the application procedure to collect information on an applicant from a third party, the requirements of the corresponding national laws must be observed. In cases of doubt, consent must be obtained from the data subject. 

There must be a legal authorization to process personal data that is related to the employment relationship but was not originally part of performance of the employment agreement. This includes legal requirements, collective regulations with employee representatives, consent of the employee, or the legitimate interest of the company.

Employee can also provide information about other people, such as employees’ dependents and families, so that the Company can provide relevant benefits or contract them in case of need. Before employee provide information to the company about other people, employee must inform them of the information they intend to provide to the company and must be responsible for the consent collection from their dependents and families. If employee share their information with the company, they may also need to read this Policy.

2.3.2 Data processing pursuant to legal authorization

The processing of personal employee data is also permitted if national legislation requests, requires or authorizes this. The type and extent of data processing must be necessary for the legally authorized data processing activity and must comply with the relevant statutory provisions. If there is some legal flexibility, the interests of the employee that merit protection must be taken into consideration.  

2.3.3 Collective agreements on data processing

If a data processing activity exceeds the purposes of fulfilling a contract, it may be permissible if authorized through a collective agreement. Collective agreements are pay scale agreements or agreements between employers and employee representatives, within the scope allowed under the relevant employment law. The agreements must cover the specific purpose of the intended data processing activity and must be drawn up within the parameters of national data protection legislation. 

2.3.4 Consent to data processing

 Employee data can be processed upon consent of the person concerned. Declarations of consent must be submitted voluntarily. Involuntary consent is void. The declaration of consent must be obtained in writing or electronically for the purposes of documentation. In certain circumstances, consent may be given verbally, in this case it must be properly documented. In the event of informed, voluntary provision of data by the relevant party, consent can be assumed if national laws do not require express consent. Before giving consent, the data subject must be informed in accordance with this Data Protection Policy.

2.3.5 Data processing pursuant to legitimate interest

2.3.6 Telecommunications and Internet

Telephone equipment, e-mail addresses, intranet, and internet along with internal social networks are provided by the company primarily for work-related assignments. They are company tools and company resources. They can be used within the applicable legal regulations and internal company policies. In the event of authorized use for private purposes, the laws on secrecy of telecommunications and the relevant national telecommunication laws must be observed if applicable.

There will be no general monitoring of telephone and e-mail communications or intranet/ internet use. To defend against attacks on the IT infrastructure or individual users, protective measures can be implemented for the connections to the FPT Smart Cloud network that block technically harmful content or that analyze the attack patterns. For security reasons, the use of telephone equipment, e-mail addresses, the intranet/internet and internal social networks can be logged for a temporary period. Evaluations of this data from a specific person can be made only in a concrete, justified case of suspected violations of laws or policies of FPT Smart Cloud. The evaluations can be conducted only by investigating departments while ensuring that the principle of proportionality is met. The relevant national laws must be observed. 

2.4. Access Request of state/government or federal agency or other regulatory body

 Requests for Personal Data Access of state/government or federal agency or other regulatory body are handled in the same way and under the same conditions as international data transfer by strictly following the requirements of the national law of the respective country. All access requests are registered in the access request register. All requests are managed by the DPO and are subject to agreement with the FPT Smart Cloud board member responsible for data protection. The DPO is responsible for communication with state/government or federal agency or other regulatory body. The DPO is responsible for the access request register. FPT Smart Cloud will inform the data subject about a request for personal data without any undue delay if it is not in contradiction to the national laws.   

2.5. Policy Review and Evaluation

 This policy must be reviewed and evaluated twice a year to reflect the latest status of international standards, legal regulations, technologies, and businesses, and to ensure the timeliness of personal data management practices.

2.6. Announce and Release

This policy is based on an announcement process that will enable personnel to understand the relevant principles and provisions of the personal data protection management policy so that they can follow it. 

This policy must be revised and reviewed by the Data Protection Officer and the responsible FPT Smart Cloud board member. The Data Protection Officer is responsible for implementation and internal audits.  

3. Data Protection Control

Compliance with the Data Protection Policy and the applicable data protection laws is checked annually with data protection audits and other controls. The performance of these controls is the responsibility of the Data Protection Representatives. The results of the data protection controls must be reported to the Data Protection Officer and the responsible FPT Smart Cloud board member. On request, the results of data protection controls will be made available to the responsible data protection authority. The responsible data protection authority can perform its own controls of compliance with the regulations of this Policy, as permitted under national law.  

4. Technical and Organizational Measures

As non-public company processing Personal Data within a scope of an agreement for commissioned data processing, the FPT Smart Cloud must take technical and organizational procedures to ensure the compliance with the European Data Protection Regulation and other international Data Protection laws. On top of such procedure, confidentiality, integrity, availability and resilience of systems and components must be guaranteed by FPT Smart Cloud.

The following groups of measures tackle all aspects of current minimum-security level. They aim at assessing FPT Smart Cloud’s level of data protection when processing personal data on behalf of the Controller. If FPT Smart Cloud connects to the Controller’s systems, FPT Smart Cloud must complete at least the confidentiality part, whereby FPT Smart Cloud will need to have the access and access authorization controls as well as the segregation of duties controls completed (sections b) c) d) below).  

Below the technical and organizational measures currently realized within FPT Smart Cloud. A continuous improvement process is implemented:

4.1. Confidentiality

a) Access Control / Building Security

The aim of the Access Control is to prevent unauthorized use of data processing systems which are used for the processing and the use of Personal Data.

Each employee’s user master data and individual identification code are registered in the contact directory. Admission to the data processing systems is only possible after identification and authentication by using the identification code and the password for the particular system.

☒ Alarm system

☒ Protection of building shafts

☒ Automatic access control system

☒ Access control by chip card transporter

☒ Locking system with code lock

☒ Manual locking system

☐ Biometric access control

☒ Video surveillance of entrances

☐ Light barriers / motion sensors

☒ Safety locks

☒ Key transfer regulation (hand-over of keys etc.)

☒ Identity check by janitor/reception

☒ Recording visitors

☒ Commitment of special selected cleaning staff

☒ Commitment of special selected security

☒ Commitment to wear authorization card staff

b) Physical Access Control/ System Protection

The aim of the Physical Access Control is to prevent unauthorised people from physically accessing such data processing equipment which processes or uses Personal Data.

Due to their respective security requirements, business premises and facilities are subdivided into different security zones with different access authorizations. They are monitored by security personnel.

Access to special security areas such as the service centre for remote maintenance or ODC is additionally protected by a separate access area. The constructional and substantive security standards comply with the security requirements for data centers.

☒ Internal access control

☒ Isolation control (permission for user rights)

☒ Strong password specification

☐ Biometric authentication

☒ Authentication a username/password

☒ Assignment of user profiles to IT Systems

☒ Locking server housing/computers

☒ Use of VPN technology (remote access)

☒ Locking external interfaces (USB etc.)

☒ Encryption of mobile data media

☒ Intrusion detection system

☐ Central smartphone administration (e.g., remote deletion)

☐ Encryption of smartphone content

☐ Secure passwords for smartphones

☒ Encryption of data media on laptop computers

☒ Assignment of individual usernames

☐ Or else, please specify:

c) Electronic Access Control/Securing Access Authorization

Measures regarding Electronic Access Control are to be targeted on the fact that only such data can be accessed for which an access authorization exists, and that Personal Data cannot be read, copied, changed, or deleted in an unauthorized manner during the processing, use and after the saving of such data.

Access to data necessary for the performance of the particular task is ensured within the systems and applications by a corresponding role and authorization concept.

 ☒ Rights authorization concept

☒ Rights management by system administrator

☒ Number of system administrators “reduced to a minimum”

☒ Recording of deletion

☒ Logging of system access events, especially entries, changes and deletions of data

☒ Application of virus protection

☒ Physical deletion of media prior to reuse

☒ Application of software firewall

☒ Secure storage of data carriers

☒ Password policies (incl. defined password length, password changes)

☒ Encryption of data carriers

☒ Use of appropriate shredders resp. specialized service providers

☒ Application of hardware firewall

☒ Proper destruction of data carriers

☐ Or else, please specify:

☒ Access logs

d) Separation control/ Measures to safeguard the separation of purposes for which Personal Data have been collected

The aim of the Separation Control is to ensure that data which have been collected for different purposes can be processed separately.

Personal Data is used by the Processor for internal purposes only. A transfer to a third party such as a Sub-Contractor is solely made under consideration of contractual arrangements and European Data Protection Regulation.

Processor’s employees are instructed to collect, process, and use Personal Data only within the framework and for the purposes of their duties (e.g., service provision). At a technical level, multi-client capability, the separation of functions as well as the separation of testing and production systems are used for this purpose.

☒ Physically separate storing using separate systems or data carrier

☒ Definition of an authorization concept

☒ Division between productive and testing systems

☒ Encryption of data records, processed for the same purpose

☒ No productive data in testing systems

☒ Logical client separation (software based)

☐ Or else, please specify:

e) Pseudonymizing

The processing of Personal Data in such a way that the data cannot be associated with a specific Data Subject without the assistance of additional information, provided that this additional information is stored separately, and is subject to appropriate technical and organizational measures.

 ☒ Pseudonymously (or anonymous) processing of data

☒ Separation of assignment file and storage in a separate, secure IT system

4.2. Integrity

 a) Data Transfer Control/Data Transfer Security

The aim of the Data Transfer Control is to ensure that Personal Data cannot be read, copied, changed, or deleted without authorization during their transfer and that it can be monitored and determined to which recipients a transfer of Personal Data is intended.

The transfer of Personal Data by FPT Smart Cloud to a third party (e.g., customers, sub-contractors, service provider) is only made if a corresponding contract exists, and only for a specific purpose. If Personal Data is transferred to companies with their seat outside the EU/EEA or the original country, FPT Smart Cloud provides that an adequate level of data protection exists at the target location or organization in accordance with the European Union’s Data Protection Regulation, e.g., by employing contracts based on the EU model contract clauses.

 ☒ Establishment of dedicated lines resp. VPN-tunnel

☒ Email encryption

☒ Recording of data recipients as well as periods of scheduled transmission resp. agreed deletion periods

☒ Physical transport: selection of special transport staff and carrier

☐ Or else, please specify:

☒ Data transfer in an anonymous or pseudonymous way

☒ Creation of an overview of regular data request as well as data transfer

☒ Physical transport: Use of secure transport containers/-packing

☒ Use of encrypted external devices when transferring data (CD, USB, stick etc.)

b) Input control

 The aim of the Input Control is to make sure with the help of appropriate measures that the circumstances of the data entry can be reviewed and monitored retroactively.

System inputs are recorded in the form of log files. By doing so, it is possible at a later stage to review whether and by whom Personal Data was entered, altered or deleted

☒ Creation of an overview proving which application entitles to input, modify or remove which data

☒ Permission settings to entitle to input, modify and delete data in accordance with a right allocation concept

☒ Continual logging of inputs, modification and deletion of data

☒ Use of individually assigned usernames to ensure access control or input, modification or deletion of data

☒ Retention of a filing system to evaluate the origin of data transmitted to automatically processed data

☒ Activity logs

☐ Or else, please specify.

4.3. Availability and Resilience

a) Availability control and protection to prevent accidental or willful destruction or loss

The aim of the availability control is to ensure that Personal Data is protected against accidental destruction and loss.

If Personal Data is no longer required for the purposes for which it was processed, it is deleted promptly. It should be noted that with each deletion, the Personal Data is only locked in the first instance and is then deleted for good with a certain delay. This is done to prevent accidental deletions or possible intentional damage.

☒ Server rooms equipped with air conditioning, protective plugs, fire extinguishers

☒ Back-ups stored separately in a safe place

☒ Emergency plan

☒ Business continuity plan

☒ No server rooms below sanitary facilities

☒ Regular data file back-ups

☒ Supervision emergency plan

☐ Or else, please specify:

b) Rapid Recovery

 ☒ Recovery acc, back-up and recovery concept

☒ Recovery testing

☒ Supervision emergency plan

4.4. Procedures to handle regular review, valuation and evaluation

a) Data Protection Management

 ☒ The principles relating to processing of personal data (collection, processing or use) are subject to an internal company policy

☒ The data protection officer has been designated in written form

☒ Employees are committed to data confidentiality/handling of personal data

☒ Employees are committed to comply with the regulations regarding the secrecy of telecommunications

☒ An internal list of processing operations is available.

☒ The data protection officer is involved in the data protection impact assessment

☒ The data protection officer is member of the organizational chart

☒ Employee training courses.

☒ Implementation of a control system designed to detect unauthorized access to personal data

☐ Or else, please specify:

b) Incident Response Management

 It corresponds to incident management in case of detected or suspected security incidents resp. failure related to IT sectors.

☒ Processing scheme for incident management

☒ Team practicing realistic exercises

☒ Security team designated and trained

☐ Or else, please specify:

c) Data protection by implementation of appropriate technical measures and privacy by default settings (as per EU Regulation)

☒ Adherence to privacy by Design/data protection by appropriate technologies

☒ Selection of privacy-enhancing technologies for future requirements

☒ Adherence to privacy by Default/data protection by appropriate settings

☐ Or else, please specify:

d) Supervision/Engagement of sub-contractors

No data processing is to be carried out without prior specific authorization of the Controller, e.g. clear contractual obligation, formalized order management, strict selection of the service provider, obligation for advance verification, follow-up inspection.

☒ Selection of (sub)contractors subject to professional diligence (in particular with regard to data security)

☒ Guidelines drawn up for processor documented in writing (e.g. by data processing agreement)

☒ Processor designated data protection officer (if necessary)

☒ Effective controller’s supervision rights agreed

☒ Prior to engagement, verification of security measures recorded by sub-contractor

☒ Processor’s employees are committed to sign a secrecy/confidentiality agreement

☒ Ensure erasure or destruction of data after termination of the contract

☒ Continuous review of processor and his activities

☐ Or else, please specify:

 5. Personal Data Protection Training

Every new employee must join the first day Personal Data Protection training.

For every employee processing personal data, it is mandatory to join the Personal Data Protection training including a successful exam before starting personal data processing. An annually refresh training is also mandatory.

6. Data Protection Officer

The Data Protection Officer, being internally independent of professional orders, works towards the compliance with national and international data protection regulations. He is responsible for the Data Protection Policy and supervises its compliance. The Data Protection Officer is appointed by the FPT Smart Cloud Board.

Any data subject may approach the Data Protection Officer, at any time to raise concerns, ask questions, request information, or make complaints relating to data protection or data security issues. If requested, concerns and complaints will be handled confidentially.

Contact details for the Data Protection Officer and staff are as follows:

FPT Smart Cloud Company, Ltd.

Data Protection Officer, Pham The Minh

FPT Tower, 10 Pham Van Bach Street, Cau Giay Ward, Ha Noi, Viet Nam 

Cell: +84 913571357

E-mail: MinhPT@fpt.com

7. Responsibilities and Disciplinary

The executive bodies of FPT Smart Cloud, subsidiaries and legal entities are responsible for data processing in their area of responsibility. Therefore, they are required to ensure that the legal requirements, and those contained in the Data Protection Policy, for data protection are met (e.g., national reporting duties). Board of Managers are responsible for ensuring that organizational, HR and technical measures are in place so that any data processing is carried out in accordance with data protection. Compliance with these requirements is the responsibility of the relevant employees. If external agencies perform data protection controls, the Data Protection Officer must be informed immediately.

Improper processing of personal data, or other violations of the data protection laws, can be criminally prosecuted in many countries, and result in claims for compensation of damage. Violations for which individual employees are responsible can lead to sanctions under employment law.

If you do not understand the implications of this policy or how it may apply to you, seek advice from the DPO via the phone or email (Pham The Minh, phone: +84913571357, email: MinhPT@fpt.com).

8. Supplementary Guidelines and Documents

Personal Data Protection Policy

Every FPT Smart Cloud employee can find these Policies, Guidelines, procedures  and templates on the platform QMS. 

9. Exceptions

Any exception must be reviewed and approved by Data Protection Officer and also approved by the responsible board member of FPT Smart Cloud.

10. Appendix

 10.1. Definition

Abbreviations

Description

 PII, Personal Identifiable Information, Personal Data

Refer to the personal data defined by the EU GDPR (Article 4 (1)), ‘personal data’ means any information relating to an identified or identifiable natural person (‘data subject’); an identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person. 

Data Subject

EU GDPR (Article 4 – 1), Data subject refers to any individual person who can be identified, directly or indirectly.

Data Controller

EU GDPR (Article 4 – 7), Data Controller means the natural or legal person, public authority, agency or anybody which alone or jointly with others, determines the purpose and means of processing of personal data; where the purpose and means of such processing are determined by Union or Member State law, the controller or the specific criteria for its nomination may be provided for by Union or Member State law.

Data Processor

EU GDPR (Article 4 – 8), Data Processor means a natural or legal person, public authority, agency or anybody which processes data on behalf of the controller.

Recipient

EU GDPR (Article 4 – 9), A natural or legal person, public authority, agency or anybody, to which the personal data are disclosed, whether third party or not.

Third Party

EU GDPR (Article 4 – 10), A natural or legal person, public authority, agency or anybody other than the data subject, controller, processor and persons who under direct authority of controller or processor, are authorized to process personal data

DPO

Data Protection Officer

DPIA

Data Protection Impacted Assessment

European Union

10.2. Related Documents

Code

Name of documents

EU GDPR

EU General Data Protection Regulation

PERSONAL DATA PROTECTION DECREE NO. 13/2023/ND-CP, VN

Decree of the Vietnamese Government: PERSONAL DATA PROTECTION DECREE NO. 13/2023/ND-CP
Nghị Định Quy Định Về Bảo Vệ Dữ Liệu Cá Nhân 07/2023

PCI DSS

Payment Card Industry Data Security Standard,

10.3. Data Protection Law, Vietnam, Overview

Data Law No. 60/2024/QH15, passed by the National Assembly on 30 November 2024. This Law comes into force as of July 1, 2025.
Criminal Code No. 100/2015/QH13, passed by the National Assembly on 27 November 2015
Law No. 24/2018/QH14 on Cybersecurity, passed by the National Assembly on 12 June 2018 (“Cybersecurity Law”);
Law No. 86/2015/QH13 on Network Information Security, passed by the National Assembly on 19 November 2015; as amended by Law No. 35/2018/QH14 dated 20 November 2018, on amendments to some articles concerning planning of 37 Laws (“Network Information Security Law”);
Law No. 59/2010/QH12 on Protection of Consumers’ Rights, passed by the National Assembly on 17 November 2010; as amended by Law No.35/2018/QH14 dated 20 November 2018, on amendments to some articles concerning planning of 37 Laws (“CRPL”);
Law No. 67/2006/QH11 on Information Technology, passed by the National Assembly on 29 June 2006; as amended by Law No. 21/2017/QH14 dated 14 November 2017 on planning (“IT Law”);
Law No. 51/2005/QH11 on E-transactions, passed by the National Assembly on 29 November 2005 (“E-transactions Law”);
Decree No. 85/2016/ND-CP dated 1 July 2016, on the security of information systems by classification (“Decree 85”);
Decree No. 72/2013/ND-CP dated 15 July 2013 of the Government, on management, provision and use of Internet services and online information; as amended by Decree No. 27/2018/ND-CP dated 1 March 2018 and Decree No.150/2018/ND-CP dated 7 November 2018 (“Decree 72”);
Decree No. 52/2013/ND-CP dated 16 May 2013 of the Government; as amended by Decree No. 08/2018/ND-CP dated 15 January 2018, on amendments to certain Decrees related to business conditions under state management of the Ministry of Industry and Trade and Decree No. 85/2021/ND-CP dated 25 September 2021 (“Decree 52”);
Decree No. 15/2020/ND-CP of the Government dated 3 February 2020 on penalties for administrative violations against regulations on postal services, telecommunications, radio frequencies, information technology and electronic transactions (“Decree 15”);
Circular No. 03/2017/TT-BTTTT of the Ministry of Information and Communications dated 24 April 2017 on guidelines for Decree 85 (“Circular 03”);
Circular No. 20/2017/TT-BTTTT dated 12 September 2017 of the Ministry of Information and Communications, providing for Regulations on coordinating and responding to information security incidents nationwide (“Circular 20”);
Circular No. 38/2016/TT-BTTTT dated 26 December 2016 of the Ministry of Information and Communications, detailing cross-border provision of public information (“Circular 38”);
Circular No. 24/2015/TT-BTTTT dated 18 August 2015 of the Ministry of Information and Communications, providing for the management and use of Internet resources, as amended by Circular No. 06/2019/TT-BTTTT dated 19 July 2019 (“Circular 25”); and
Decision No. 05/2017/QD-TTg of the Prime Minister dated 16 March 2017 on emergency response plans to ensure national cyber-information security (“Decision 05” ).

FPT Corporation Data protection Regulation:

 Vietnamese: Chinh sach bao mat du lieu ca nhan (01-CS/TT/HDCV/FPT v1.0) Chính sách bảo mật dữ liệu cá nhân

 Vietnamese: Chinh sach bao mat du lieu ca nhan cua can bo nhan vien (02-CS/TT/HDCV/FPT v1.0) Chính sách bảo mật dữ liệu cá nhân của cán bộ nhân viên

Vulnerability Disclosure | FPT Smart Cloud

Report a security or privacy vulnerability

If you believe you have discovered a security or privacy vulnerability in an FPT Smart Cloud product, please report it to us.

I. How to report a security or privacy vulnerability

If you believe you have discovered a security or privacy vulnerability that affects FPT Smart Cloud products, software, services, or web servers, please report it to us. We welcome reports from everyone, including security researchers, developers, and customers.

To report a security or privacy vulnerability, please send an email to security_report@fpt.com that includes:
• The specific product and software version(s) which you believe are affected
• A description of the behavior you observed as well as the behavior that you expected
• A numbered list of steps required to reproduce the issue and a video demonstration, if the steps may be hard to follow

Please encrypt sensitive information that you send by email. You’ll receive a reply from FPT Smart Cloud to acknowledge that we received your report, and we’ll contact you if we need more information.

II. How FPT Smart Cloud handles these reports

For the protection of our customers, FPT Smart Cloud doesn’t disclose, discuss, or confirm security issues until our investigation is complete and any necessary updates are generally available.

FPT Smart Cloud uses security advisories and our security-announce mailing list to publish information about security fixes in our products and to publicly credit people or organizations that have reported security issues to us.

For more information on CVD, please review the information provided in the following links:
https://www.iso.org/standard/72311.html

Security Advisories

FPT Smart Cloud Security Advisories are a supplement to the FPT Smart Cloud Security bulletins. They address security changes that may not require a security bulletin but that may still affect customers’ overall security.

FPT Smart Cloud Security Advisories are a way for FPT Smart Cloud to communicate security information to customers about issues that may not be classified as vulnerabilities and may not require a security bulletin. Each advisory is accompanied by an FPT Smart Cloud Knowledge Base Article to provide additional information about any changes or updates being delivered with the advisory’s release.

Help protect your computing environment by keeping up to date on FPT Smart Cloud technical security notifications. For more information, see FPT Smart Cloud Technical Security Notifications.

Artificial Intelligence (AI) is reshaping our world at a rapid pace. From optimizing business processes to supporting critical decisions, AI offers infinite potential, fostering a synergy between AI and humans. However, alongside its immense benefits, AI also poses significant challenges, especially concerning ethics. The question is no longer “How will AI develop?” but rather “How should AI develop responsibly?”.

This is where AI ethics becomes a timely and urgent topic, not only for technology developers but also for regulators, businesses, and society as a whole. Let’s explore what AI ethics is with FPT.AI!

What is AI Ethics?

AI ethics is a set of principles, values, and guidelines that aim to ensure the transparent, fair, and responsible development, deployment, and use of artificial intelligence. According to UNESCO, AI ethics doesn’t just focus on what the technology can do, but also revolves around the questions of “What should AI do?” and “How should we use AI to benefit humanity, rather than cause harm?”.

Why is AI Ethics Important?

AI is no longer merely supporting technology. It’s a powerful tool capable of making decisions. According to Deputy Minister of Science and Technology, Mr. Bui The Duy, AI is completely different from any other technologies that humans have found before. While old technological products only follow existing instructions, AI can create its own directions, beyond the control of developers.

Furthermore, AI has now become an indispensable partner for humans. A global study involving approximately 32,000 workers from 47 countries by The University of Melbourne showed that over 58% of employees actively use AI in their work, with one-third of them using AI weekly or daily. .

Ảnh được thiết kế bởi Yevgenia Nayberg

Therefore, without ethical oversight, sometimes wrong decisions made by AI can have far-reaching impacts on individuals or even a nation. An algorithm could reject a job application simply because their resume came from a rural area, implicitly assuming they are “less promising.” A facial recognition system could misidentify people of color due to a lack of diverse data. A chatbot could learn discriminatory language from social media users if left unchecked.

These examples highlight that if AI lacks ethics, the consequences will not be limited to technical errors but will also lead to social, legal, and humanitarian consequences.

Core Values Shaping AI for the Future

In the journey of artificial intelligence development, what’s important is not just how far technology advances, but which direction we are leading it. To ensure AI serves the common good, for people, society, and this planet, UNESCO has outlined four core values that act as guiding principles, including:

Respecting human rights and human dignity: Ensuring the respect, protection, and promotion of human rights, fundamental freedoms, and the dignity of each individual.

Building peaceful, just, and interconnected societies: Encouraging the development of societies where everyone can live in harmony, fairness, and connection.

Promoting diversity and inclusion: AI must be designed to serve everyone, excluding no one, fostering diversity and creating equal opportunities.

Protecting the environment and developing thriving ecosystems: AI technology needs to be environmentally responsible, contributing to the protection of the planet and natural ecosystems.

These values are crucial compasses for guiding AI development in a positive, sustainable direction.

Core Principles in AI Ethics

With the goal of “living safely” with AI, philosopher Luciano Floridi has distilled 5 core principles, becoming reliable beacons on the journey alongside AI:

Beneficence: AI should be developed and applied to improve the lives of humans and our planet. The ultimate goal is to create “AI for Social Good” (AI4SG), where AI is used to enhance societal well-being.

Nonmaleficence: The principle of “do no harm” is paramount, especially when AI has the potential to affect human existence. This includes avoiding harm to privacy, autonomy, and employment opportunities.

Autonomy: Human ability to act freely and independently must be preserved and promoted, while machine autonomy needs to be limited.

Justice: AI must be developed, designed, and deployed in a way that promotes justice, fairness, equality, and related values. This requires addressing issues like algorithmic bias and ensuring equitable access.

Explicability: To promote other principles, we need to understand the “how” and “why” behind AI systems and products. Accountability and transparency are key.

What Should Businesses Do?

For businesses, investing in AI technology cannot be separated from building an ethical foundation. According to Coursera, many global corporations like IBM, Google, and Microsoft have established internal ethics committees, built codes of conduct, and verification processes to ensure their AI products adhere to ethical standards from the outset.

Some specific actions regarding AI ethics that businesses can take include:

Training personnel on AI ethics, data, and privacy.

Integrating ethical risk assessment into the product development process.

Consulting independent experts to review critical algorithms.

Being transparent about how data is collected, processed, and used.

Establishing internal ethics councils to verify products before launch.

Not Just One Person’s Responsibility

AI ethics is not solely a matter for the tech industry; it’s a shared responsibility of society as a whole:

Nations and government agencies need to create flexible, appropriate, and practical legal frameworks that keep pace with technological development while still safeguarding human rights.

Universities and research institutes need to integrate ethics topics into AI and data science curricula.

AI users also need to enhance their understanding to use AI intelligently and responsibly.

Towards a Sustainable Technological Future

AI ethics is not a barrier to innovation, but a solid foundation for the development of artificial intelligence. By prioritizing these core values, we are building a future where AI is not only exceptionally intelligent but also deeply humane, serving and enhancing human life.

As AI becomes increasingly integrated into decision-making systems, AI ethics is no longer an option – it’s a prerequisite for building a just, transparent, and humane digital society.

🤝 Build ethical AI today, create lasting trust tomorrow.

Sources

UNESCO. (n.d.). Recommendation on the Ethics of Artificial Intelligence. https://www.unesco.org/en/artificial-intelligence/recommendation-ethics

Coursera. (2023, July 26). What is AI ethics? Definition and examples. https://www.coursera.org/articles/ai-ethics

IBM. (n.d.). AI ethics. IBM. https://www.ibm.com/think/topics/ai-ethics

University of Texas at Austin. (n.d.). AI ethics. Ethics Unwrapped. https://ethicsunwrapped.utexas.edu/glossary/ai-ethics thics Unwrapped. https://ethicsunwrapped.utexas.edu/glossary/ai-ethics

In the advanced age of artificial intelligence, every breakthrough technology marks a pivotal shift — and the latest buzz driving the tech world is the Model Context Protocol.

Launched in late 2024 by Anthropic, the parent company of Claude, Model Context Protocol (MCP) has rapidly gained global attention. This new protocol is designed to revolutionize the way AI Agents and Large Language Models (LLMs) interact, enabling seamless communication and unlocking collaboration at an unprecedented scale.

But what exactly is MCP? How is it being applied in real-world scenarios? And is it related to the familiar APIs we use today? Join FPT.AI as we dive into the fascinating world of Model Context Protocol and uncover the value it brings to the future of AI collaboration.

What is MCP?

MCP, or Model Context Protocol, is essentially a set of rules and standards that enables different AI Agents and Large Language Models (LLMs) to exchange information, share resources, and coordinate on complex tasks. MCP acts as the “USB-C” of the AI world — a universal connector that allows AI models to plug into tools, services, and external data sources, resulting in more accurate and context-aware responses for real-world business needs.

Figure 1 – What is MCP?

MCP Architecture

To operate seamlessly, Model Context Protocol is built on three core components:

Figure 2 – MCP Architecture

MCP Host – A chatbot, IDE, or AI-powered tool that acts as the central coordinator. It manages sessions, controls access, and can initiate commands to MCP based on user requests or automated workflows.
MCP Client – A web or mobile application initiated by the Host. It connects to a single Server and facilitates two-way communication between the Host and the Server.
MCP Server – Acts as a bridge to external tools or data sources (e.g., Google Drive, Slack), enabling functions such as file retrieval, status updates, and data access:
- Prompt – Predefined instructions for LLMs, easily triggered via slash commands (/search), menus, or interactive UI elements.
- Source – Structured data (files, databases, histories) that provides essential context to AI models.
- Tool – Functional units that empower AI agents to take action, such as calling an API or writing data.

Why MCP Matters?

Beyond just a connection protocol, MCP brings a host of powerful advantages that enhance flexibility, security, and scalability across AI systems:

Standardized Communication: MCP provides a structured framework that enables AI models to interact with diverse tools in a consistent and unified manner.

Real-time Tool Access & Integration: AI assistants can now leverage external tools to gather real-time data — boosting operational efficiency and reducing costs for businesses.

Enterprise-grade Security & Scalability: MCP enables secure and seamless integration with enterprise applications, supporting agile growth without compromising data protection.

Multi-modal Integration: MCP supports multiple communication methods — including STDIO, SSE (Server-Sent Events), and WebSocket — ensuring compatibility across different deployment environments. This makes it easy for businesses to expand their AI ecosystems simply by connecting new MCP Servers, without disrupting existing workflows.

API vs MCP: Key Similarities and Differences

To implement MCP effectively, businesses need to understand how it differs from traditional APIs.

Figure 3. Differences between MCP and API

When to Use API vs MCP

On the one hand, business can use APIs when precision, predictability, and strict control are paramount in a tightly scoped environment.
For example:

Online Banking Applications – Tasks like checking balances or transferring funds demand high security and accuracy, with clearly defined operations.

On the other hand, businesses will use MCP when you need dynamic, intelligent coordination between AI Agents and LLMs for complex, multi-step tasks.
For example:

Business Trip Planning – Instead of integrating with calendar, flight, and email services separately, an AI Agent can interact with all via MCP in one seamless flow.

Smart IDEs – MCP simplifies connections with file systems and version control, enabling AI to better understand code context and make smarter suggestions.

Complex Data Analysis – AI platforms can automatically discover and interact with data sources and visualization tools via a unified MCP layer.

A Major Leap Toward “Build Your Own AI Agents” journey

Staying ahead of the curve, FPT AI Agents has integrated MCP Client Tools into its platform. This game-changing feature allows your AI Agents to connect and interact with external tools via MCP Servers — unlocking new levels of flexibility and intelligence.

With this integration, users can:

Declare secure endpoints with flexible auth options (No Auth / Bearer Token).

Select or exclude specific tools from connected MCP Servers as needed.

Embed directly into business processes, allowing AI Agents to automate smart workflows powered by external capabilities.

This upgrade empowers enterprise AI teams to become more agile, versatile, and impactful — enabling smarter collaboration across systems and unlocking greater business value.

The Future of MCP in the AI-Driven World

Although still emerging, MCP is already reshaping the way we interact with and scale artificial intelligence:

Standardizing AI Communication – Like HTTP standardized the web, MCP is poised to become the universal language for AI agents and tools to talk, collaborate, and co-create.

Enabling a New AI Economy – MCP fosters a thriving ecosystem where enterprises and indie developers alike can build and expand AI-powered products.

Accelerating the Evolution of AI Agents – With a consistent interaction framework, MCP makes intelligent, autonomous AI agents a practical reality.

Redefining Human-AI Collaboration – By allowing machines to understand and respond in more humanlike, contextual ways, MCP moves us closer to true augmented intelligence.

Conclusion

MCP is more than just a protocol — it’s a dynamic gateway that enables AI tools to understand deeper context and deliver smarter, more accurate responses. With its unmatched advantages in scalability, flexibility, and performance, MCP is set to become a core trend shaping the future of AI.

At FPT AI Agents, the adoption of MCP Client Tools empowers your business to harness this breakthrough — enabling smarter automation, cost savings, and enhanced productivity.

Let’s build better, smarter AI Agents together — with FPT.AI!

Sources:

Base.vn. (n.d.). Model Context Protocol (MCP) là gì?. https://base.vn/blog/model-context-protocol-mcp-la-gi/

BitOnTree. (2024, March 19). Model Context Protocol vs API: What’s the difference and why it matters. https://www.bitontree.com/blog/model-context-protocol-vs-api

Microsoft Tech Community. (2024, April 23). Unleashing the power of Model Context Protocol (MCP): A game-changer in AI integration. https://techcommunity.microsoft.com/blog/educatordeveloperblog/unleashing-the-power-of-model-context-protocol-mcp-a-game-changer-in-ai-integrat/4397564

AI Agents are artificial intelligence systems that can interact with the environment and make decisions to achieve goals in the real world without any human guidance or intervention. This technology are shaping technology trends, with notable milestones such as the Google I/O 2023 event launching Astra or the emergence of GPT-4o.

Large corporations are pouring billions of dollars into AI Agents to take the lead in AI Era. In this article, FPT.AI will clarify how AI Agents are helping businesses improve processes, enhance customer experience and optimize operations.

What are AI Agents (Intelligent Agents)?

AI Agents are artificial intelligence systems that can interact with the environment and make decisions in the real world without any human guidance or intervention.

AI Agents can gather information from their surroundings, design their own workflows, use available tools, coordinate between different systems, and even work with other Agents to achieve goals without requiring user supervision or continuous new instructions.

With the development of Generative AI, Natural language processing, Foundation Models, and Large Language Models (LLMs), AI Agents can now simultaneously process multiple types of multimodal information such as text, voice, video, audio, and code. Advanced agent AI can learn and update their behavior over time, continuously experimenting with new solutions to problems until achieving optimal results. Notably, they can detect their own errors and find ways to correct them as they progress.

AI Agents can exist in the physical world (robots, autonomous drones, or self-driving cars) or operate within computers and software to complete digital tasks. The aspects, components, and interfaces of each agent AI can vary depending on its specific purpose. Encouragingly, even people without deep technical backgrounds can now build and use AI Agents through user-friendly platforms.

AI Agents are AI models and algorithms that are capable of making decisions without human intervention.

>>> READ NOW: What is Generative AI? Trends in Applying GenAI from 2024 to 2027

What are the key features of an AI Agent platform?

Key features of an AI Agent platform include:

Autonomy: AI Agents can operate independently, make decisions, and take actions without continuous human supervision. For example, self-driving cars can adjust speed, change lanes, stop, or adjust routes based on real-time sensor data about road conditions and obstacles, without driver intervention.
Reasoning Ability: AI agents use logic and analyze available information to draw conclusions and solve problems. They can identify patterns in data, evaluate evidence, and make decisions based on the current context, similar to human thinking processes.
Continuous Learning: AI Agents continuously improve their performance over time by learning from data and adapting to changes in the environment. For instance, customer support chatbots can analyze millions of conversations to gain deeper understanding of common issues and improve the quality of proposed solutions.
Environmental Observation: AI agents continuously collect and process information from their surroundings through techniques like computer vision, natural language processing, and sensor data analysis. This ability helps them understand the current context and make appropriate decisions.
Action Capability: AI agents can perform specific actions to achieve goals. These actions can be physical (like a robot moving objects) or digital (like sending emails, updating data, or triggering automated processes).
Strategic Planning: AI agents can develop detailed plans to achieve goals, including identifying necessary steps, evaluating alternatives, and selecting optimal solutions. This ability requires predicting future outcomes and considering potential obstacles.
Proactivity and Reactivity: AI agents proactively anticipate and prepare for future changes. For example, Nest Thermostat learns the homeowner’s heating habits and proactively adjusts temperature before the user returns home, while quickly responding to unusual temperature fluctuations.
Collaboration Ability: AI agents can work effectively with humans and other agents to achieve common goals. This collaboration requires clear communication, coordinated actions, and understanding the roles and objectives of other participants in the system.
Self-Improvement: Advanced AI agents can self-evaluate and improve their operational performance. They analyze the results of previous actions, adjust strategies based on feedback, and continuously enhance their capabilities through machine learning techniques and optimization.

Key Features of AI Agents

>>> READ NOW: 2 Ways to Classify Artificial Intelligence and 7 Common Types of AI

Differences between Agentic AI Chatbots and AI Chatbots

Below is a comparison table highlighting the distinctions between Agentic AI chatbots and AI Chatbots:

Criteria	Agentic AI Chatbots	Traditional AI Chatbots
Autonomy	Operate independently, perform complex tasks without continuous intervention	Require continuous guidance from users, only respond when prompted
Memory	Maintain long-term memory between sessions, remember user interactions and preferences	Limited or no memory storage capability, each session typically starts from scratch
Tool Integration	Use function calls to connect with APIs, databases, and external applications	Operate in closed environments with no ability to access external tools or data sources
Task Processing	Break down complex tasks into subtasks, execute them sequentially to achieve goals	Only process simple, individual requests without ability to decompose complex problems
Knowledge Sources	Combine existing knowledge with new information from external sources (RAG)	Rely solely on pre-trained data, unable to update with new information
Learning Capability	Continuously learn from interactions, improving accuracy and relevance over time	Do not learn or improve from user interactions, responses always follow fixed patterns
Operation Mode	Can perform multiple processing rounds for a single request, creating multi-step workflows	Operate on a single-turn basis (receive-process-respond), without multi-step capabilities
Planning Ability	Strategically plan and self-adjust when encountering new information or obstacles	No long-term planning capability or strategy adjustment
Personalization	Provide personalized experiences based on user history, preferences, and context	Deliver generalized responses, identical for all users
Response Process	Analyze intent, access relevant information, create plan, execute actions, and evaluate results	Recognize patterns, search for appropriate responses in existing database, reply
Error Handling	Recognize errors, self-correct, and find alternative solutions when problems arise	Often fail to recognize errors or lack ability to recover when encountering off-script situations
User Interaction	Proactively ask clarifying questions, suggest options, and track progress	Passive, only directly respond to what users explicitly ask
Workflow	Use threads to store all information, connect with tools, execute function calls when needed	Simple processing according to predefined scripts, no workflow extension capability
Practical Applications	Complex customer support, data analysis, process automation, personal assistance	Primarily for FAQs, basic customer support, simple conversations
Intent Detection	Accurately identify users’ underlying intents, even when not explicitly stated	Only react to specific keywords or patterns, often missing true intentions
System Integration	Easily integrate with multiple systems and applications through APIs	Limited integration capabilities, often requiring custom solutions
Development Requirements	Can be developed on no-code platforms, without requiring in-depth programming knowledge	Typically require programming knowledge to build and maintain

Agentic AI chatbots mark a significant evolution in conversational AI, powered by LLMs but extending well beyond them. Operating on thread-based architecture, they store complete conversation histories, files, and function call results. These advanced chatbots activate via various triggers (scheduled events, database changes, or manual inputs) to analyze requests, interpret intentions, and execute actions autonomously.

Five key innovations drive this technology:

RAG integration for context-aware responses with higher accuracy
Function calling to interact with external systems
Advanced memory systems for continuous learning and adaptation
Tool evaluation to assess resources and fill information gaps
Subtask generation to break down complex goals independently

Unlike traditional chatbots’ single-turn model (receive-process-respond), agentic chatbots process multiple turns per prompt, queue actions strategically, and dynamically select appropriate tools based on user intent. They can search connected knowledge bases, call external APIs, or generate responses from core training when external tools aren’t needed. Critically, no-code platforms have democratized their development, accelerating adoption across industries by enabling businesses of all sizes to implement sophisticated AI without significant technical investment.

Differences between Agentic AI chatbots and AI Chatbots

>>> READ MORE: What is Agentic RAG? Difference between Agentic RAG and RAG

Key Components of AI Agents

AI Agents are composed of multiple components working together as a unified system, similar to how the human body functions with senses, muscles, and brain. Each component in AI Agent Architecture plays a specific role in helping the agent sense, think, and interact with the surrounding world.

Key components of AI Agents

Sensors

Sensors help AI Agents collect information (percepts) from the surrounding environment to understand the context and current situation. In physical robots, sensors might be cameras for “seeing,” microphones for “hearing,” or thermal sensors for “feeling” temperature. For software agents running on computers, sensors might be web search functions to gather online information, or file reading tools to process data from PDF documents, CSV files, or other formats.

Sensors help AI Agents collect information (percepts) from the surrounding environment

>>> EXPLORE MORE: How to build an AI Agent and train it successfully?

Actuators

If sensors are how agents receive information, actuators are how they affect the world. Actuators are components that allow agents to perform specific actions after making decisions. In physical robots, actuators might be wheels for movement, mechanical arms for lifting objects, or speakers for producing sound. For software agents, actuators might be the ability to create new files, send emails, control other applications, or modify data in systems.

Actuators are components that allow agents to perform specific actions after making decisions

Brain

Processors, Control Systems, and Decision-Making Mechanisms form the “brain” of the AI Agents, where information is processed and decisions are made. Processors analyze raw data from sensors and convert it into meaningful information. Control systems coordinate the agent’s activities, ensuring all parts work harmoniously. Decision-making mechanisms are the most important part, where the agent “thinks” about processed information, evaluates different action options, and selects the most optimal action based on goals and existing knowledge.

Processors, Control Systems, and Decision-Making Mechanisms form the “brain” of the AI Agent

>>> EXPLORE: Applications of AI Agents in Personalized Marketing

Learning and Knowledge Base Systems

These are the memory and learning capabilities of AI Agents, allowing them to improve performance over time. Knowledge base systems store information the agent already knows: data about the world, rules of action, and experiences from previous interactions. This might be a database of locations, events, or problems the agent has encountered along with corresponding solutions.

Learning systems allow the agent to learn from experience, recognize patterns, and improve decision-making abilities. An agent with learning capabilities will continuously update its knowledge base, helping it better cope with new situations or changes in the environment.

The complexity level of these components depends on the tasks the AI Agent performs. A smart thermostat might only need simple temperature sensors, a basic control system, and actuators to turn heating systems on/off. In contrast, a self-driving car needs to be equipped with all components at high complexity levels: diverse sensors to observe roads and other vehicles, powerful processors to handle large amounts of real-time data, sophisticated decision-making systems for safe navigation, precise actuators to control the vehicle, and continuous learning systems to improve driving capabilities through each experience.

AI Knowledge Management Agents

>>> EXPLORE: What Are Intelligent Agents? The Difference Between AI Agents and Intelligent Agents

How do AI Agents Work?

When receiving a command (goal) from a user (Prompt), AI Agents immediately initiate the goal analysis process, transferring the prompt to the core AI model (typically a Large Language Model) and beginning to plan actions. The Agent will break down complex goals into specific tasks and subtasks, with clear priorities and dependencies. For simple tasks, the Agent may skip the planning stage and directly improve responses through an iterative process.

During implementation, thanks to Sensors, AI agents collect information (transaction data, customer interaction history) from various sources (including external datasets, web searches, APIs, and even other agents). During this collection process, the AI Agent continuously updates its knowledge base, self-adjusts, and corrects errors if necessary.

The Processors of AI Agents use algorithms, Deep Neural Networks, machine learning models, and artificial intelligence to analyze information and calculate necessary actions.

Throughout this process, the agent’s Memory continuously stores information (such as history of decisions made or rules learned). Additionally, AI Agents also use feedback from users, feedback from other Agents, and Human-in-the-loop (HITL) to self-compare, adjust, and improve performance over time, avoiding repetition of the same errors.

Finally, through Actuators, AI Agents perform actions based on their decisions. For robots, actuators might be parts that help them move or manipulate objects. For software agents, this might be sending information or executing commands on systems.

Technically, an AI agent system consists of four main components, simulating the way humans operate

To illustrate this process, imagine a user planning their vacation. They ask an AI Agent to predict which week of the coming year will have the best weather for surfing in Greece. Since the large language model that underpins the agent is not specialized in weather forecasting, the agent must access an external database that contains daily weather reports in Greece over the past several years.

Even with historical data, the agent cannot yet determine the optimal weather conditions for surfing. Therefore, it must communicate with a surf agent to learn that ideal surfing conditions include high tides, sunny weather, and low or no rainfall.

With the newly gathered information, the agent combines and analyzes the data to identify relevant weather patterns. Based on this, it predicts which week of the coming year in Greece is most likely to have high tides, sunny weather, and low rainfall. The final result is then presented to the user.

According to BCG analysis, AI agents are strongly penetrating many business processes, with a compound annual growth rate of up to 45% over the next 5 years

>>> READ NOW: RPA vs AI Agents: Is RPA Still Relevant in the Age of AI?

Common Types of AI Agents

There are 5 primary types of AI Agents: Simple Reflex Agents, Goal-Based AI Agents, Model-Based Reflex Agents, Utility-Based Agents, Learning Agents. Each suited to specific tasks and applications:

Simple Reflex Agents: Simple Reflex Agents operate on the “condition-action” principle and respond to their environment based on simple pre-programmed rules, such as a thermostat that turns on the heating system at exactly 8pm every night. The agent does not retain any memory, does not interact with other agents without information, and cannot react appropriately if faced with unexpected situations.
Model-Based Reflex Agents: Model-Based Reflex Agents use their cognitive abilities and memory to create an internal model of the world around them. By storing information in memory, these agents can operate effectively in changing environments but are still constrained by pre-programmed rules. For example, a robot vacuum cleaner can sense obstacles when cleaning a room and adjust its path to avoid collisions. It also remembers areas it has cleaned to avoid unnecessary repetition.
Goal-Based AI Agents: Goal-Based Agents are driven by one or more specific goals. They look for appropriate courses of action to achieve the goal and plan ahead before executing them. For example, when a navigation system suggests the fastest route to your destination, it analyzes different paths to find the most optimal one. If the system detects a faster route, it updates and suggests an alternative route.
Utility-Based Agents: Utility-Based Agents evaluate the outcomes of decisions in situations with multiple viable paths. They employ utility functions to measure the usefulness that each action might bring. Evaluation criteria typically include progress toward goals, time requirements, or implementation complexity. This evaluation system helps identify the ideal choice: Is the best option the cheapest? The fastest? The most efficient? For example, a navigation system considers factors such as fuel economy, reduced travel time, and toll costs to select and recommend the most favorable route for the user.
Learning Agents: Learning Agents learn through concepts and sensors, while utilizing feedback from the environment or users to improve performance over time. New experiences are automatically added to the Learning Agent’s initial knowledge base, helping the agent operate effectively in unfamiliar environments. For example, e-commerce websites use Learning Agents to track user activity and preferences, then recommend suitable products and services. The learning cycle repeats each time new recommendations are made, and user activities are continuously stored for learning purposes, helping Agents improve the accuracy of their suggestions over time.

Popular Types of AI Agents

>>> EXPLORE: What is an LLM Agent? How it works, advantages, and disadvantages

What are the outstanding benefits of using AI Agents?

AI Agents for businesses deliver a consistent experience to customers across multiple channels, with the following 4 outstanding benefits:

Improve productivity: AI Agents help automate repetitive and time-intensive tasks, freeing up human resources from manual work so that businesses can focus on more strategic, creative and high-value initiatives, fostering innovation. For more complex issues, AI Agents can intelligently escalate cases to human agents. This seamless collaboration ensures smooth operations, even during periods of high demand.
Reduce costs: By optimizing processes and minimizing human errors, AI personnel help businesses cut operating costs. Complex tasks are handled efficiently by AI Agents without the need for constant human intervention.
Make informed decisions: AI Agents use machine learning (ML) technologies to help managers collect and analyze data (product demand or market trends) in real time, making faster and more accurate decisions.
Improve customer experience: AI agents significantly enhance customer satisfaction and loyalty by offering round-the-clock support and personalized interactions. Their prompt and precise responses effectively address customer needs, ensuring a smooth and engaging service experience. Lenovo leveraged AI agents to streamline product configuration and customer service, integrating them into key systems like inventory tracking. By building a knowledge database from purchase data, product details, and customer profiles, AI agents help Lenovo cut setup time from 12 minutes to 2 minutes, boosting sales productivity and customer experience. This led to a 12% improvement in order delivery KPIs (within 17 days) and generated $5.88 million in one year, according to Gartner.

Benefits of implementing AI Agents in Business

>>> Read more about: AI Agents at Work – Foundation for Productivity Breakthrough

Is ChatGPT an AI Agent?

ChatGPT is not an AI Agent. It is a large language model (LLM) designed to generate human-like responses based on received input, with some components similar to AI Agents:

Simple sensors that receive text input
Actuators that generate text, images, or audio
Control system based on transformer architecture
Knowledge base system from pre-training data and fine-tuning.

However, these elements are not sufficient to make ChatGPT a genuine Agent. The most important difference between AI Agents and ChatGPT is autonomy. ChatGPT cannot set its own goals, make plans, or take independent actions. When you ask ChatGPT to write an email, it can create content but cannot send the email itself or evaluate whether sending an email is the best action in a specific situation.

Additionally, ChatGPT cannot directly interact with external systems or adjust its behavior based on real-time feedback. Updates like plugins, extended frameworks, APIs, and prompt engineering can improve ChatGPT’s functionality, but still don’t create a complete Agent. ChatGPT also lacks the ability to maintain long-term memory between sessions. It doesn’t “remember” you or previous conversations unless specifically programmed to do so in certain applications.

ChatGPT lacks core features to be considered an AI Agent

>>> READ NOW: What is a Multi Agent System (MAS)?

Practical Applications of AI Agents

Imagine a future workplace where every employee, manager, and leader not only works together, but is also equipped with a team of AI teammates to support them in every task and at every moment of the workday. With these AI teammates, we will become 10x more productive, achieve better results, create higher quality products, and of course, become 10x more creative.

You may be wondering, “When will this future come?” The answer from FPT is: The future is now. Here are four stories that demonstrate how AI is already impacting businesses.

Revolutionizing Insurance Claims Processing

Imagine you go to the hospital for a health check-up, buy medicine, and file an insurance claim. Typically, the insurance company’s document processing will take at least 20 minutes. With integrated AI Agents, insurers can process all documents through rapid assessment tools, risk assessment tools, and fraud detection tools, returning results in just 2 minutes.

This represents an incredible leap in productivity, improving the customer experience and creating new competitive value for the business.

AI Agents in Finance – Accounting

>>> READ NOW: Blockchain, Deepseek & AI Agents Reshape the AI Race

Transforming the Customer Contact Center

The second story focuses on customer service. Several FPT.AI customers have deployed AI systems for inbound and outbound communications. These systems provide human-like customer support, handling requests, resolving issues, and providing excellent service.

For some customers, AI Agents are now handling 70% of customer requests, completing 95% of received tasks, and achieving a customer satisfaction rating of 4.5/5. Currently, FPT’s customer service AI Agents manage 200 million user interactions per month.

How AI Agents Improve Customer Service

Advantages of applying AI Agents in customer service

Empowering pharmacists with AI Mentor

At Long Chau, the largest pharmacy chain in Vietnam, more than 14,000 pharmacists work every day to advise customers. To ensure they stay updated with knowledge and work effectively, FPT.AI has developed an AI Mentor that interacts with more than 16,000 pharmacists across 2,000 pharmacies every day.

This AI Mentor identifies strengths and weaknesses, provides insights, and personalizes conversations to help them improve. The results are:

Pharmacists’ competencies improved by 15%.
Productivity increased by 30%.

Within the first nine months of the year, the pharmacy chain recorded a revenue growth of 62%, reaching VND 18.006 trillion, accounting for 62% of FRT’s total revenue and completing 85% of its 2024 plan. More importantly, we pride ourselves on helping pharmacists become the best versions of themselves while continuously improving.

FPT AI Mentor won the “Outstanding Artificial Intelligence Solution” award at AI Awards 2024

>>> READ NOW: Understanding AI Agents in KYC

From a cost center to a profit center

FPT.AI’s AI Innovation Lab works with customers to identify opportunities, deploy pilots, and scale solutions. For example, one of our clients transformed their customer service center from a cost center to a profit center.

Using AI, they detected when customers were happy and immediately suggested appropriate products or services to upsell credit cards, cross-sell overdrafts, activate new customers to sign up, and reactivate existing customers. This approach helped the customer service center contribute about 6% of total revenue.

The four stories above are just a small part of the countless ways AI can transform businesses. AI, as a new competitive factor, is opening up a blue ocean of innovation. Every company and organization will need to reinvent their operations and build a strong foundation to compete in the future, leveraging the advances of AI.

Applications of AI Agents in practice

>>> EXPLORE: What is Agentic AI? The differences between Generative AI and Agentic AI

Challenges in Deploying AI Agents

AI Agents are still in their early stages of development and face many major challenges. According to Kanjun Qiu, CEO and founder of AI research startup Imbue, the development of AI Agents today can be compared to the race to develop self-driving cars 10 years ago. Although AI Agents can perform many tasks, they are still not reliable enough and cannot operate completely autonomously.

One of the biggest problems that AI Agents face is the limitation of logical thinking. According to Qiu, although AI programming tools can generate code, they often write wrong or cannot test their own code. This requires constant human intervention to perfect the process.
Dr. Fan also commented that at present, we have not achieved an AI Agent that can fully automate daily repetitive tasks. The system still has the ability to “go crazy” and not always follow the exact user request.

Challenges and Considerations When Using AI Agents

Another major limitation is the context window – the ability of AI models to read, understand, and process large amounts of data. Dr. Fan explains that models like ChatGPT can be programmed, but have difficulty processing long and complex code, while humans can easily follow hundreds of lines of code without difficulty.
Companies like Google have had to improve the ability to handle context in their AI models, such as with the Gemini model, to improve performance and accuracy.

For “physical” AI Agents such as robots or virtual characters in games, training them to perform human-like tasks is also a challenge. Currently, training data for these systems is very limited and research is just beginning to explore how to apply generative AI to automation.

>>>> EXPLORE: What is Data Leakage? How to Prevent Data Leakage when implementing Generative AI?

Continue writing the future with AI Agents with FPT.AI

In the digital economy, competition between companies and countries is no longer based solely on core resources, technology and expertise. Organizations, from now on, will need to compete with a new important factor: AI Companions or AI Agents.

It is expected that by the end of 2025, there will be about 100,000 AI Agents accompanying businesses in customer care, operations and production. Each AI Agent will undertake a number of tasks such as programming, training, customer care… Thanks to that, employees are more empowered, businesses increase operational productivity, improve customer experience, and make more accurate decisions based on data analysis.

The Future of AI Agents

FPT AI Agents – a platform that allows businesses to develop, build and operate AI Agents in the simplest, most convenient and fastest way. The main advantages of FPT AI Agents include:

Easy to operate and use natural language.
Flexible integration with enterprise knowledge sources.
AI models are optimized for each task and language.

Currently, FPT AI Agents supports 4 languages: English, Vietnamese, Japanese and Indonesian. In particular, AI Agents have the ability to self-learn and improve over time.

FPT AI Agents is FPT Smart Cloud’s trump card in the AI era

AI Agents are all operated on FPT AI Factory – an ecosystem established with the mission of empowering every organization and individual to build their own AI solutions, using their data, supplementing their knowledge and adapting to their culture. This differentiation fosters a completely new competitive edge among enterprises and extends to building AI sovereignty among nations.

FPT AI Agents Deployment Process

With more than 80 cloud services and 20 AI products, FPT AI Factory helps accelerate AI applications by 9 times thanks to the use of the latest generation GPUs, such as H100 and H200, while saving up to 45% in costs. These factories are fully compatible with the NVIDIA AI Enterprise platform and architectural blueprints, ensuring seamless integration and operation.

>>> READ NOW: Why Gen AI Agents are the future prospect of Generative AI?

FAQs about AI Agents

What’s the difference between LLMs and AI Agents?

LLM (Large Language Model) is an AI model trained on a vast amount of data to recognize and generate natural language. It functions as a “language brain,” predicting each next word in a sentence. However, traditional LLMs are limited to their initial training data, lack the ability to interact with the outside world, and cannot update themselves with new information after training.

AI Agent, on the other hand, is a much more complete system that typically uses an LLM as its core intelligence foundation but is supplemented with sensors (gathering information), actuators (performing actions), knowledge bases (storing knowledge), and control systems (making decisions). This structure allows AI Agents to not only understand language but also interact with the surrounding environment.

The decisive difference between LLMs and Agents lies in the AI Agent’s “tool calling” capability. Through this mechanism, an Agent can retrieve updated information from external sources, optimize workflows, automatically break down and solve subtasks, store interactions in long-term memory, and plan for future actions. These capabilities help AI Agents provide more personalized experiences and comprehensive responses, while expanding their practical application across many fields.

>>> EXPLORE: Are AI Agents the next frontier of NLP Chatbots?

Are Reasoning models (like OpenAI o3 and DeepSeek R1) AI agents?

No. Reasoning models like o3 and R1 are LLMs trained to reason through solutions to complex problems. They do this by breaking them down into multiple steps using chain of thought. These LLMs cannot naturally interact with other systems or extend their reasoning beyond architectural limitations.

How do AI Agents integrate with existing systems and workflows?

The most common ways to integrate AI agents are:

Connect it to a RAG platform, providing native connections between LLMs and knowledge bases. This allows the agent to use representations of your documents and business data as context for future responses, increasing output accuracy.
Through APIs to external services. When you configure function calls in the AI agent platform, the model interacts with API endpoints in the same way a traditional program would, creating all the headers and body of the call.

How does Human-in-the-loop fit into the AI Agent workflow?

Human-in-the-loop frameworks enhance supervision of AI agent systems. Simply put, the agent’s actions pause at predetermined points in the workflow. A notification is sent to the user, who must review decisions, information, and scheduled tasks. Based on this information, the user will approve or change how the AI agent will continue the task.

Will AI agents take our jobs?

This technology will certainly replace jobs and bring changes to the market, although there is no clear vision of when and how this might happen. Workers may be replaced by AI agents in many industries. At the same time, many positions for AI development and maintenance may be created, along with human-in-the-loop positions, to ensure that human decisions control AI actions rather than the other way around.

Do AI agents exacerbate bias and discrimination?

An AI model is only as unbiased as the data it was trained on—so yes, they are biased. Addressing these issues involves changing machine learning processes and creating datasets that represent the full spectrum of the world and human experiences.

Who is responsible when an AI Agents makes a mistake?

A difficult problem in ethics and law, it’s still unclear who should be blamed for accidents and unintended consequences. The developers? Hardware/software owners? Operators? As new laws are created and industry barriers are implemented, we will be able to understand what roles AI agents can—and cannot—assume.

In short, with the ability to be autonomous, operate independently, make decisions based on data and real-world environments, AI Agents are a powerful automation solution that helps businesses optimize processes. The AI Agents market is forecast to reach a size of 30 billion USD by 2033 and maintain a growth rate of about 31% per year.

The explosive potential of this technology in the next decade is huge. Contact FPT.AI now to take advantage of the enormous power of AI colleagues, accelerate innovation, enhance customer experience and scale more efficiently than ever!

>>> Read more about:

Agentic RAG is a method that combines the power of Retrieval-Augmented Generation (RAG) with AI Agents, creating intelligent, proactive, and flexible information retrieval and generation systems. Compared to traditional RAG, Agentic RAG can actively determine when, how, and what needs to be retrieved from various diverse data sources.

In this article, FPT.AI will introduce in detail the nature, operating mechanism, and comprehensive differences of Agentic RAG compared to traditional RAG. Through this, readers will clearly understand the potential as well as the limitations of this technology, thereby making the right decisions when choosing solutions suitable for specific business needs. References are also listed at the end of our articles (in case you would like to look for deeper insights).

What is Agentic RAG?

Agentic RAG is a method that combines the power of Retrieval-Augmented Generation (RAG) with AI Agents to enhance content creation and decision-making capabilities in artificial intelligence systems. While traditional RAG systems supplement large language models with information from external sources according to fixed retrieval strategies, Agentic RAG can actively decide which information is relevant, which information should be prioritized, and how to adjust the content creation process to suit contexts or needs that change in real-time.

Agentic RAG combines the power of RAG with AI Agents to enhance content creation and decision making

Agentic RAG opens new potential for AI applications that require both accurate information retrieval and complex decision-making. By combining the power of RAG and AI Agents, Agentic RAG not only enhances the quality of retrieved information but also optimizes how that information is used in the content creation process.

How Agentic RAG Works

>>> EXPLORE: How to build an AI Agent and train it successfully?

How does Agentic RAG Work?

Unlike traditional RAG, which uses Retrievers and Generators operating separately, Agentic RAG integrates one or more types of AI Agents into the RAG system (Multi-Agent Framework). These AI Agents collaborate to process complex queries together.

For example, an Agentic RAG system can combine multiple information retrieval Agents, each specializing in a specific domain or type of data source. For instance, one Agent might focus on querying External Databases while another searches emails or web results. This task allocation creates a high level of specialization in the information processing.

Single-Agent RAG (Router)

>>> EXPLORE: What is a Multi Agent System (MAS)?

Single-Agent RAG (Router)

An Agentic RAG system can include AI Agent types such as:

Routing Agents: Routing Agents determine which knowledge sources and tools will be used to process user queries. They process prompts and select the appropriate RAG pipeline to create optimal responses. In single-Agent RAG systems, the Routing Agent will select the data source that needs to be retrieved.
Query planning Agents: Query planning agents act as task managers in the RAG pipeline. They break complex queries into smaller steps and distribute them to other agents. After receiving results from specialized Agents, Query Planning Agents combine the responses into a comprehensive, complete result. This mechanism, called AI Orchestration, allows the system to efficiently process complex multi-dimensional queries.
ReAct Agents: ReAct (reasoning and action) is an agent framework that helps create multi-agent systems capable of reasoning and acting step by step. Notably, ReAct Agents can determine the appropriate tool for each specific task. Based on step-by-step results, ReAct agents can flexibly adjust subsequent steps.
Plan-and-execute Agents: This is an advanced version of ReAct agents that can perform multi-step processes without needing to return to the primary agent. This mechanism helps reduce processing costs and increase system efficiency. Since this Agent must develop a comprehensive plan from the beginning, the task completion rate and result quality are usually higher than other Agent types.

ReAct framework in Agentic RAG

Frameworks that can be found on GitHub such as LangChain, LlamaIndex, and the LangGraph Orchestration Framework help simplify the implementation of Agentic RAG. Using open-source models like Granite™ or Llama-3 also helps reduce costs and increase observability.

RAG Multi-Agent System

>>> EXPLORE: What is an LLM Agent? How it works, advantages, and disadvantages

What is RAG?

Retrieval Augmented Generation is an artificial intelligence (AI) technique that enhances the performance of large language models (LLMs) by connecting Generative AI models with an External Knowledge Base. Instead of relying solely on available training data, RAG helps AI models access real-time data through APIs and other connections to data sources.

How Retrieval Augmented Generation Works

A standard RAG pipeline consists of two main components:

Information retrieval component (Retriever): Typically an Embedding Model combined with a Vector Database containing data to be retrieved. Retrievers usually search for information relevant to the input query in huge datasets or document repositories.
Generation component (Generator): Usually an LLM like GPT, BERT, or similar architectures. The Generator processes the query and retrieved documents to create coherent and contextually appropriate responses.

When receiving a natural language query, the Embedding Model converts the query into a Vector Embedding, then retrieves similar data from the Knowledge Base. The AI system combines the retrieved data with the user query to create contextually appropriate responses.

A standard RAG pipeline

The main advantage of RAG lies in its ability to reference updated information or specialized data that may not have been included in the model’s training phase. This minimizes the hallucination problem, where language models provide information that seems reasonable but is inaccurate, while ensuring higher factual accuracy. RAG allows LLMs to operate more accurately in specialized contexts without needing fine-tuning.

RAG is widely applied in fields requiring accuracy and contextual relevance in content creation such as:

Customer support: RAG provides accurate responses by retrieving relevant information from product manuals, FAQs, or customer databases.
Medicine and research: RAG enhances language models to create deep insights by retrieving and referencing academic articles or research datasets.
AI Chatbots: Specialized chatbots are significantly improved by RAG, ensuring that responses are informed by a broader dataset than what was used in the initial training process.

Difference between Pipeline RAG and Agentic RAG

>>> EXPLORE: RPA vs AI Agents: Is RPA Still Relevant in the Age of AI?

What are AI Agents?

AI Agents are types of AI that can interact with the environment, process input information, and perform a sequence of actions based on specified inputs or goals without human intervention. Most current Agents are large language models (LLMs) with Function Calling capabilities, meaning they can call tools to perform tasks.

Key Components of AI Agents

The main roles of Agents are to automate tasks, optimize processes, and make intelligent decisions in dynamic environments, particularly suitable for complex decision-making tasks. Theoretically, AI Agents are LLMs with three prominent characteristics:

Possessing both short-term and long-term memory, able to reference previous tasks to plan and execute complex subsequent tasks.
Having the ability to route queries, plan step by step, and make decisions. AI Agents have memorization capabilities to retain information and outline actions appropriate for complex queries.
Having the ability to call tools through APIs. More advanced Agents can even actively choose appropriate tools to optimize the user response process.

The Agent workflow (Agentic Workflow) can include a single AI Agent or a system of multiple Agents working together. Agents can vary in complexity, from simple rule-based systems to complex models leveraging Deep Learning.

AI Agents are a type of AI that can automatically interact with the environment, process input information and perform actions

Based on characteristics and functions, AI Agents can be classified into several groups. Reactive Agents operate based on the current state of the environment, following predetermined rules or responses without storing or using past experiences.

Cognitive Agents are more advanced with the ability to store past experiences, analyze patterns, and make decisions based on memory, often used in systems requiring learning from previous interactions. Collaborative Agents interact with other Agents or systems to achieve common goals, commonly found in multi-Agent systems where multiple Agents collaborate, share information, or coordinate actions.

In terms of architecture and communication, Agents rely on various architectures, including decision-making models, Neural Networks, and rule-based systems. Communication between Agents is typically conducted through protocols such as message passing, event triggering, or interactions based on complex networks, particularly important in distributed systems.

Agents can be organized according to centralized models, where all decisions are made by a single controlling entity, or distributed, where each Agent operates autonomously but still contributes to a larger goal.

Main Types of AI Agents

>>> EXPLORE: Applications of AI Agents in Personalized Marketing

Differences between Agentic RAG and Traditional RAG

See the detailed comparison table between Agentic RAG and traditional RAG:

Criteria	Traditional RAG	Agentic RAG
Operating mechanism	Passive information retrieval, only when requested	Adds a decision-making layer through autonomous Agents, actively decides when, how, and what needs to be retrieved
Flexibility	Connects LLM with a single dataset	Can retrieve data from multiple External Knowledge Bases and use external tools
Adaptability	Reactive data retrieval tool, does not adapt to changing contexts, requires prompt engineering to achieve optimal results	Solves problems intelligently and flexibly, Agents coordinate and check each other
Accuracy	Does not self-verify or optimize results	Can iterate the process to optimize results over time
Scalability	Limited due to connection with a single data source	Higher thanks to a network of Agents working together, accessing multiple data sources, and using Tool-Calling
Multimodality	Usually limited to text processing	Leverages Multimodal LLMs to process diverse data such as images and audio
Cost	Lower due to using fewer tokens	Higher because it needs more Agents and tokens
Latency	Lower	Higher because LLMs need time to generate responses
Reliability	Depends on the quality of source data	May fail depending on complexity and type of Agent used

Thus, the most fundamental difference between Agentic RAG and traditional RAG lies in proactivity and decision-making ability. Traditional RAG operates as a passive tool, only retrieving information when requested and based on a rigid process established in advance. In contrast, Agentic RAG integrates intelligent Agents capable of actively deciding the process of searching, processing, and synthesizing information.

While traditional RAG is like an employee strictly following given instructions, Agentic RAG operates like a team of autonomous experts, not only performing assigned tasks but also having the ability to store and reference previous query sets, contexts, and results (through Semantic Caching), analyze problems, coordinate with each other, and provide creative solutions.

Summary of the differences between RAG, AI Agents and Agentic RAG

However, Agentic RAG is not always better than traditional RAG. Having multiple AI Agents means higher costs, as more tokens are needed. Additionally, LLMs can create latency because they take time to generate responses. Moreover, Agentic RAG still fails in complex tasks, competing for resources, leading to conflicts. And even the best RAG systems cannot completely eliminate the possibility of “hallucination.”

Therefore, businesses should only choose Agentic RAG when they need to solve complex problems requiring multiple data sources, need high flexibility in searching and processing information, or want systems capable of self-improving accuracy over time. With limited budgets, needing quick response solutions with simple tasks and clearly defined data sources, traditional RAG remains an effective and cost-efficient choice.

Distinguishing between AI, Agentic AI, RAG and Agentic RAG

>>> EXPLORE: What is Agentic AI? The differences between GenAI and Agentic AI

Notable Applications of Agentic RAG

Agentic RAG can be used in most applications of traditional RAG, but due to higher computational demands, it is more suitable in situations requiring queries across multiple data sources. Some applications include:

Real-time question answering and decision support: In situations requiring rapid data analysis such as stock market analysis or medical diagnosis, businesses deploy AI chatbots, virtual assistants, or FAQ systems using RAG to provide accurate, updated information to employees and customers.
Automated support: With the ability to retrieve content relevant to ongoing conversations and automate customer service with personalized and contextually appropriate content, businesses can use Agentic RAG to handle simple support requests and forward more complex issues to human staff.
Data management: RAG systems help quickly retrieve information in internal databases, reducing employees’ manual search needs.
Multi-Agent collaboration systems: Agentic RAG shows great potential in distributed AI systems where multiple Agents need to coordinate work on large datasets or process complex queries, creating an intelligent network with superior information processing capabilities.

Outstanding applications of Agentic RAG

In conclusion, Agentic RAG marks a significant advancement in artificial intelligence by combining the power of retrieval-generation and intelligent multi-Agent systems. The choice between Agentic RAG and traditional RAG needs to be carefully considered based on specific requirements, available resources, and the complexity of the task.

In the future, with the continuous development of large language models and Agent technology, Agentic RAG promises to become increasingly refined, overcoming current limitations and expanding its application scope in various fields of life and business.

References:

Weaviate. (n.d.). What is Agentic RAG. Retrieved April 20, 2025, from https://weaviate.io/blog/what-is-agentic-rag
IBM. (n.d.). What is Agentic RAG? Retrieved April 20, 2025, from https://www.ibm.com/think/topics/agentic-rag
LeewayHertz. (n.d.). Agentic RAG: What it is, its types, applications and implementation. Retrieved April 20, 2025, from https://www.leewayhertz.com/agentic-rag/

>>> EXPLORE:

AI Chatbot is a computer program that uses artificial intelligence to simulate human-to-human conversation, capable of understanding and responding to user requests naturally and accurately. In this article, FPT.AI will present in-depth information about the core technologies, types of AI Chatbot, and their popular applications. The article also provides detailed reviews of the features and advantages of the top 9 AI chatbot online free that you can leverage to level up your performance.

What is an AI Chatbot?

AI Chatbot is a computer program designed to simulate human conversation, interacting with users through text or voice. AI Chatbot combine artificial intelligence, Machine Learning, and Natural Language Processing (NLP) to understand and analyze user requests, then apply Natural Language Generation (NLG) techniques to provide appropriate, accurate, and natural human-like responses.

How AI Chatbot Works

In reality, you may have already interacted with chatbots through various activities, from smart speakers in your home, popular messaging applications like SMS, WhatsApp, and Facebook Messenger, to work platforms like Slack. These bots simulate natural conversation and try to resolve your queries before transferring to human representatives if necessary.

With the ability to automate conversations, AI Chatbot help organizations and businesses flexibly scale, personalize, and improve communication capabilities in many activities – from internal workflows and customer service to DevOps management.

The best AI chatbots of 2025

Review of TOP 9 AI Chatbot online free

AI Chatbot	Developer	Service Plans	Key Strengths	Special Features	Best For
FPT AI Chat	FPT Smart Cloud (Vietnam)	Business service package, contact for detailed consultation	Superior Vietnamese language processing (>90% accuracy); Multi-platform integration	Rating Agent; User Reaction; Typing Suggestion	Vietnamese businesses seeking customer service automation
ChatGPT	OpenAI	Free to $200/month	Globally popular; Continuously updated features	DALL-E 3 (image generation); Deep Research; Advanced Voice Mode; Sora (AI video)	Individuals and businesses needing comprehensive AI solutions
Claude	Anthropic	Free to $20/month	Large context memory (150,000 words); Pleasant conversational style	Artifacts (interactive data dashboards); Computer use (experimental)	Long document analysis and natural conversations
Deepseek	DeepSeek (China)	Completely free (open source)	Strong logical reasoning; Can run on personal computers	DeepSeek R1 comparable to ChatGPT o3	Developers and small organizations needing free AI
Microsoft Copilot	Microsoft	Free to paid	Deep integration with Microsoft Office	Unlimited image generation; Personalized experience	Microsoft Office users (Word, Excel, PowerPoint)
Gemini	Google	Free	Integration with the entire Google ecosystem	Direct Python code writing/execution; Gems (customizable like GPTs)	Gmail, Google Drive and Google services users
Meta AI	Meta	Completely free	Direct integration with Meta social networks	Developed on LLaMA 3 model (70 billion parameters), can create animated images and videos	Facebook, Instagram, Messenger and WhatsApp users
Perplexity	Perplexity AI	Free to $20/month	Results displayed with source citations	Purpose-specific searches; Conversation sharing	Research and information searches requiring clear sources
Grok	xAI (Elon Musk)	Completely free	Real-time updates; Humorous style	DeepSearch; Brainstorm; Analyze Data; Create Images	Users needing quick information updates and creativity

FPT.AI Chat

FPT.AI Chat is one of the leading AI chatbot platforms in Vietnam, developed by FPT Smart Cloud – a member of FPT Technology Group. This artificial intelligence solution is designed to help businesses automate customer care, optimize consultation processes, and increase business efficiency. Not only possessing superior Vietnamese natural language processing technology, FPT.AI Chat also stands out with its deep and flexible integration capabilities, suitable for many different industries.

The difference between FPT AI Chat and ChatGPT

FPT.AI Chat uses Generative AI technology and Deep Learning, helping the chatbot correctly understand the intent and context of user questions. The accurate response rate can reach over 90%. This allows the chatbot to quickly and accurately handle common questions related to products, prices, promotions, or purchasing processes, thereby reducing repetitive work for consultants, helping them focus on situations requiring higher expertise.

Additionally, FPT.AI Chat has multi-platform integration capabilities, from websites and mobile applications to popular channels such as Facebook Messenger, Zalo, Viber, Instagram, WhatsApp, and even the business social network GapoWork. Businesses only need to build a chatbot once and can deploy it across multiple channels, effectively expanding customer reach.

Besides the main features, FPT.AI Chat is continuously updated with add-ons based on actual needs:

Rating Agent: Allows customers to rate their satisfaction with consultants or with the chatbot itself after each conversation session, helping businesses measure support quality and improve services.
User Reaction: Users can respond with emoji reactions to chatbot answers, creating a more natural and interesting conversation experience. These interactions are compiled to improve response quality.
Typing Suggestion: Suggests questions for users based on keywords being typed, saving time and directing the conversation more effectively.
Copy Model: Useful in the testing and deployment phases of chatbots, allowing the copying of model structures from a test bot to an official bot, ensuring consistency and accuracy when put into actual operation.

Since July 2024, LPBank has officially deployed FPT.AI Chat on their website, Zalo OA, and fanpages to automate 24/7 customer care processes. This banking chatbot can handle a range of common issues such as account registration, product information, interest rate consultation… and is ready to connect with consultants when customers need in-depth support. This brings a seamless, modern experience that aligns with the trend of digitizing banking services.

According to representatives from FPT Smart Cloud and LPBank, the FPT.AI chatbot is not just a support tool but an intelligent interaction bridge, helping the bank maintain engagement with customers across multiple channels. This is a powerful “made in Vietnam” AI chatbot solution, suitable for Vietnamese businesses looking to effectively digitally transform, enhance customer care quality, and optimize operations with advanced AI technology.

LPBank has deployed FPT AI Chat to improve customer service

ChatGPT

ChatGPT, developed by OpenAI, has become a global phenomenon and is currently the most popular free AI chat software. This AI Chatbot uses Deep Learning technology to generate natural, intelligent, and contextually accurate responses.

Users can ask questions and communicate via text, request tasks such as writing paragraphs, writing programming code, and composing emails as if conversing with a real person. ChatGPT also supports many different languages, including Vietnamese, ensuring it meets the learning and working needs of Vietnamese users.

ChatGPT provides many outstanding features such as:

Content management
Programming support
Processing and analyzing big data to provide detailed reports
Predicting business trends and supporting strategic decision-making
Creating images from text through DALL-E 3
Analyzing and summarizing data

Factors that make up ChatGPT’s superior language processing capabilities

With the latest version, OpenAI has added many creative features such as:

Search tool to find information on the web and cite sources
Deep Research acts as an assistant that reads and analyzes all search results to create in-depth reports
Projects allows uploading documents and setting up system guidelines to direct how ChatGPT responds
Canvas is a new output mode that helps users write together with ChatGPT
Advanced Voice Mode allows real-time voice interaction
Operator AI Agent can perform tasks by browsing the web
Sora, OpenAI’s AI video model (currently only available to Pro users in the United States)

Regarding service packages, ChatGPT offers several options:

ChatGPT 3.5: Free, but with limited uses per day
ChatGPT Plus: $20/month, unlocks many advanced features and unlimited use
ChatGPT Pro: $200/month
ChatGPT Team: $20-30/user/month, designed for small teams
ChatGPT Enterprise: Variable pricing depending on needs

To use ChatGPT, users only need to access the website https://chatgpt.com/ and log in with an existing account or register a new one. When starting a conversation, you just need to enter content in the box at the bottom of the screen. When results are displayed, a new item will be created in the left menu, helping you easily manage and interact with previous conversations.

However, the biggest limitations of the free version of ChatGPT that you should note include:

Limited usage per day, which can cause interruptions in your workflow.
Does not provide source citations and has no indicators of the accuracy of answers.

Compare ChatGPT versions

Claude

Claude AI is an artificial intelligence integrated chatbot developed by Anthropic – a technology startup based in San Francisco, USA, notable for its direction of developing AI that is “safe, ethical, and transparent.” Although still unfamiliar to most Vietnamese users, Claude AI is increasingly highly rated and considered one of the smartest chatbots available today, supporting users with:

Answering questions
Creating content based on requests
Writing emails
Translating between multiple languages, including Vietnamese
Summarizing long text files and website content
Processing images or scanned documents, recognizing handwritten text with high accuracy, even when the content is in Vietnamese

A superior advantage of Claude is its ability to remember large contexts – up to 150,000 words in a conversation. This is ideal for uploading long documents like PDF files, analyzing and discussing them in depth without “forgetting” information like many other chatbots. Claude also saves conversation history, allowing users to review previous queries – very convenient for those who frequently analyze data or work with repetitive information.

Claude AI’s Core Competencies

Especially recently, this AI Chatbot launched the “Artifacts” feature, allowing users to create visual, interactive data dashboards that can be edited right in the conversation, such as personal budget planning tables or simple physics games with just a few command lines. Everything is updated in real-time without having to leave the chat interface.

Claude currently has three AI model versions, each suitable for different needs:

Claude Haiku: Fast, concise, suitable for user-oriented applications and big data processing.
Claude Sonnet: More advanced, capable of reasoning and writing complex code. This version is currently testing the “computer use” feature, allowing AI to operate like a real user on a computer.
Claude Opus: The strongest, excelling in natural language understanding, content creation, and programming. Although stronger than Sonnet, the processing speed is still comparable.

Performance comparison between versions of Claude AI

Claude is often rated as having a more pleasant, emotional conversational style compared to ChatGPT. Therefore, this AI Chatbot is very suitable for business environments or when communicating with customers. Additionally, Claude also has an active community on Reddit, where users share experiences, discuss, and support each other during use.

To use Claude AI, users only need to access the website claude.ai/login to register or log in with a Gmail account. It’s worth noting that in the free package, the number of messages per day from Claude may vary depending on overall usage levels. If you want to experience all features, you can upgrade to the Pro package for $20/month, unlocking additional utilities such as integration with Zapier to automate workflows.

The free version of Claude has limited uses

Deepseek

DeepSeek is an artificial intelligence model developed in China, with two main versions: DeepSeek V3 and DeepSeek R1. Of these, DeepSeek R1 is the latest version, specially designed to handle tasks related to logical reasoning and problem-solving. Many users rate DeepSeek R1 as comparable to powerful OpenAI models like ChatGPT o3.

Comparing DeepSeek v3 with ChatGPT and Claude

Notably, DeepSeek R1 is an open-source AI Chatbot and completely free to use. Users can access the model through a web application or download it to run on personal computers or servers, as long as the device is powerful enough to process it. This opens up many opportunities for AI access to individuals, small organizations, or developers wanting to integrate AI without licensing costs.

The interface of the DeepSeek application – whether on the web or mobile devices – is quite basic, focusing only on the main function of text input. Users can send questions, request web searches, upload text documents for content extraction, and review the history of previous conversations.

However, DeepSeek doesn’t understand images, only reading the text portions of documents. Additionally, another notable disadvantage to consider is the data security issue. Because DeepSeek application servers are located in China, the storage and processing methods of user data remain unclear. This may be concerning for those who prioritize personal information privacy.

DeepSeek User Interface

Microsoft Copilot

Built on OpenAI’s AI platform, Microsoft Copilot is not just an AI Chatbot but a comprehensive virtual assistant, helping to optimize workflows, increase productivity, and deliver a seamless user experience for both individuals and businesses. The greatest strength of Microsoft Copilot is its deep integration into Microsoft Office flagship products such as Word, Excel, PowerPoint, Teams, the Edge browser, and the Windows 11 operating system.

In Word, you can ask Copilot to draft text from simple suggestions; in Excel, the tool can analyze data and create charts; and in PowerPoint, it helps build professional presentation slides with just a few lines of guidance. This is a major advantage over standalone chatbots like ChatGPT, especially for users already familiar with the Microsoft work environment.

Users can quickly access it from the taskbar to ask questions, search the Internet, suggest ideas, or quickly open installed applications without downloading additional software. If you don’t see the Copilot icon, simply search for the term “Copilot” in the Windows search menu to activate it.

Copilot has powerful integration with Microsoft Office work management tools

Not only supporting office work, Copilot also impresses with its ability to create unlimited images, serving well for illustration needs, simple design, or creating visual content. This is an outstanding point compared to many current AI chatbots that don’t support image creation.

Another interesting aspect of Copilot is its ability to customize and personalize the user experience. On first use, Copilot may ask your name to personalize the conversation. Responses from Copilot not only answer questions but often end with expansion suggestions, inviting users to continue interacting. Additionally, Microsoft’s large user community, from official forums to Reddit groups, provides a helpful and dynamic support ecosystem.

Microsoft offers two versions: a free version with basic features and Copilot Pro, allowing access to the latest AI models, prioritized processing, and integration of more in-depth features. Furthermore, Microsoft is developing specialized versions for specific fields such as sales, finance, or cybersecurity – suitable for large organizations wanting to use AI to improve operational efficiency.

However, some users feel that responses from Copilot are sometimes not deep enough or lack updated elements such as analyzing modern trends (e.g., changes due to digitization, streaming…). Additionally, compared to competitors like ChatGPT, some versions of Copilot are rated as not “sharp” enough, feeling like a “condensed” version.

Microsoft Copilot user interface

Gemini

Gemini is an AI chatbot developed by Google with the ability to deeply integrate with the entire Google ecosystem – from Gmail, Drive, Docs, Sheets to services like Maps, Flights, or YouTube. Users only need to log in once to access it directly from Gmail, Calendar, or Docs without switching between platforms, maximizing the use of familiar tools in work and daily life. This transforms Gemini into a truly intelligent assistant tied to Google usage habits.

You can ask Gemini to search for emails in your Gmail inbox, summarize document content in Drive, check flight or hotel prices in real-time, and even plan vacations and suggest packing lists for trips.

Beyond convenience, Gemini is also highly rated for its ability to understand deep context and remember long conversation information (this AI Chatbot can store content equivalent to the entire Harry Potter series). This allows Gemini to process complex queries, analyze in-depth content, and respond naturally, like having a real conversation with a knowledgeable person.

Google Gemini User Interface

Not limited to text, Gemini can generate multimedia content such as videos, images, and music. Additionally, users can write and run Python code directly within the interface – a very useful feature for programmers or highly technical users.

Another special feature is Gems – a feature that allows personalizing the experience similar to GPTs in ChatGPT. Users can create customized Gemini versions by adding specific instructions for the chatbot to respond according to their own style or purpose.

However, Gemini’s response quality is sometimes inconsistent – with the same query, you might receive different answers if you try again later. Therefore, when using it for important tasks, users should carefully check information before applying it. Google is also transparent about this, displaying warnings that Gemini may make mistakes and encouraging users to verify content.

Comparing Microsoft Copilot, ChatGPT 4o and Google Gemini

Meta AI

Meta AI is an AI chatbot software developed by Meta – the parent company of Facebook – and integrated directly into familiar applications such as Facebook, Instagram, Messenger, and WhatsApp. The introduction of Meta AI marks a step forward in bringing artificial intelligence closer to social media users by leveraging the very platforms that millions of people use daily.

Meta AI uses the LLaMA 3 language model, the latest version, with over 70 billion parameters. This is one of the most powerful large language models currently available, capable of understanding and processing language well and supporting users in various tasks such as chatting, content creation, information searching, and even creating images, short videos, or animations – completely free.

Llama 3’s superior performance across a variety of tasks

Meta AI also supports Vietnamese language, making it very friendly and accessible to Vietnamese users. Thanks to direct integration into Facebook, Instagram, and Messenger, you only need to log in with your social media account to start chatting with AI right within the application you’re using, without downloading additional apps or switching platforms.

Meta AI also has the ability to create images and animations right in the conversation. Users can request AI to create illustrative images based on descriptions, even simple short videos.

Meta AI still has some weaknesses:

Web information search capability is not as strong as ChatGPT or Gemini; users need to verify information from sources for important content
Limited extensibility, not yet suitable for deep customization or expansion to specialized business needs.

However, if you are a developer, Meta offers a great opportunity with its open licensing policy for LLaMA models. You can use this model to build your own AI applications without paying licensing fees, as long as revenue doesn’t exceed a very high threshold. This helps individuals and small businesses easily access powerful AI technology without financial barriers.

Meta AI is integrated directly into Facebook, Instagram and Messenger, users just need to log in with their social network account to start using it

Perplexity

Perplexity is an AI chatbot notable for its ability to conduct in-depth information searches and provide clear source citations. Unlike traditional AI tools that often just provide answers, Perplexity is an intelligent search tool that can display search results with a list of sources, helping users access, verify, and continue to explore content easily and visually.

You can continue expanding your search by entering additional questions or selecting from a list of related search suggestions. Responses will be stacked in a scrolling format, allowing you to easily follow the information flow throughout the conversation.

Perplexity also supports purpose-specific search modes, such as product suggestions, healthy cooking recipes, or finding hotels through TripAdvisor. You just need to check “Copilot” in the search bar, and this AI Chatbot will present a series of sub-questions to clarify your intention, then compile the most accurate and complete content.

Perplexity User Interface

Perplexity also has the ability to share conversations: recipients can continue the conversation from where you left off, while you can see the number of views, likes, and follow-up questions from the community. This AI Chatbot also has a “Discover” section – a place that compiles popular searches into short, easy-to-understand, and engaging articles. Technologically, Perplexity uses many powerful AI models such as OpenAI GPT, Claude, and DeepSeek, while also leveraging Wolfram Alpha to process real-time data or solve complex problems.

The free version of Perplexity allows users to use some basic features, including 3 Pro queries per day. Meanwhile, the Perplexity Pro package ($20/month) expands access to up to 300 Pro queries/day, along with the ability to analyze files, display data in image format, and choose AI models at will.

Regarding the interface, Perplexity is rated as clean and easy to use, with help sections, guides, and shortcuts clearly arranged. The system always operates stably, and the user community on Discord as well as Perplexity’s blog regularly share knowledge and effective usage tips.

Comparing Perplexity and ChatGPT

Grok

Grok is an AI chatbot developed by xAI, Elon Musk’s company. The name “Grok” comes from an English word meaning “to understand intuitively,” and true to its name, Grok was built to become an AI Chatbot with deep reasoning abilities and intelligent responses.

Initially, Grok was only available within the X social media platform (formerly Twitter) and required users to pay. However, currently, you can use Grok for free through web and mobile applications, making it more accessible to many people. The latest version of this AI Chatbot is Grok 3, released in February 2025, quickly entering the top chatbot rankings, surpassing many other well-known competitors.

Grok is an AI chatbot developed by xAI, Elon Musk’s company

This AI Chatbot has a diverse system of functions, with options serving many learning, work, and entertainment needs such as:

DeepSearch: Dive deep into a specific topic
Brainstorm: Support creative idea generation
Analyze Data: Aanalyze documents, data
Create Images: Create images on request (for example, “an otter playing ukulele”)
Code: Programming support

Regarding response style, Grok has a friendly approach with a touch of light humor, not too serious but also not excessively comedic. Some users describe Grok’s style as “slightly rebellious” compared to GPT-4o, but overall it maintains coherence and professionalism when needed.

Another significant advantage of Grok is its ability to update and respond in real-time. Because it’s deeply integrated with the X platform – which frequently updates trends and news – Grok can answer questions about hot topics quickly and flexibly.

However, Grok has also been involved in controversies related to accuracy, especially when handling topics with political elements. This requires users to verify information in sensitive cases or those of high importance.

Comparing the response times of Grok, ChatGPT and DeepSeek

Outstanding benefits of AI Chatbot

AI Chatbot brings clear benefits to both businesses and customers:

24/7 Customer Support: Before chatbots, all customer inquiries required direct human response. This was especially difficult when issues arose outside working hours, on weekends, or holidays. Maintaining a 24/7 support team requires high costs and is challenging to manage. With AI Chatbot, businesses can solve human resource issues, ensuring timely support, eliminating waiting times, and ensuring customers always receive immediate responses.
Reduced Operating Costs: Implementing chatbots can help save up to 30% of operating costs, reducing the need for customer support center staffing, while minimizing time and costs related to training employees to answer repetitive queries.
Enhanced Customer Experience and Loyalty:The ability of chatbots to respond quickly and accurately creates an excellent user experience. Satisfied customers are more likely to show loyalty to the brand and complete purchase transactions.
Improved Human Resource Management: Chatbots automate workflows and free employees from repetitive tasks, allowing them to focus on more complex issues. Businesses can enhance staff efficiency to meet increased demands.

Benefits of AI Chatbot for Business

Generate Leads and Increase Conversion Rates: Chatbots can assist customers in the purchasing process by answering questions about products or services on the spot, helping customers move toward making a purchase. For complex transactions, chatbots can assess lead quality and connect them with sales agents.
Continuous Learning and Improvement: With Deep Learning technology and Natural Language Processing (NLP), AI chatbot continuously learns from previous interactions to improve response quality, becoming smarter and more flexible in handling complex queries.
Internal Administration Support: AI Chatbot integrates into internal management systems, helping automatically retrieve and synthesize data from digital document repositories. Departments such as human resources, accounting, or operations management can quickly access information, create reports, and make decisions based on real-time data.
Multi-channel Integration and Technology Ecosystem: Chatbots easily integrate with many communication platforms such as websites, social media, and mobile applications. This flexibility helps businesses reach customers at various touchpoints while creating a seamless technology ecosystem to effectively collect and analyze customer data.

Benefits of AI Chatbots for Users

Notable Applications of AI Chatbot

Common use cases for chatbots include:

Personalized E-commerce: In online retail, e-commerce chatbots provide personalized product recommendations based on purchase history and browsing behavior, helping increase conversion rates and improve shopping experiences.
Conversational Marketing Strategies: Marketers use Marketing chatbots to promote products, services, and collect insights about customer interaction and purchasing patterns, creating engaging and personalized conversations across web and messaging channels.
Office Process Automation: In finance and healthcare fields, AI chatbot helps identify fields in forms, capture customer information, and schedule appointments for healthcare offices, minimizing processing time and errors.
Employee Self-service: IT and HR teams use chatbots to allow employees to solve common issues themselves, such as resetting passwords, accessing company policy information, or asking benefits-related questions.
Intelligent Call Management: At contact centers, chatbots function as Interactive Voice Response (IVR) systems, streamlining incoming communications and directing customers to the right resources, along with the ability to seamlessly transfer to support staff when necessary.
Smart Personal Assistants: Chatbots can provide automatic reminders for time or location-based tasks, helping users manage schedules and work more effectively.
Advanced Conversational Analytics: Chatbot technology also helps extract valuable information from natural language communications between customers and businesses, providing deep insights to improve services and products.

Chatbot interaction interfaces are also very diverse, from social media messaging applications, independent messaging platforms, website integration and proprietary applications, to phone calls. This helps businesses easily reach customers across multiple channels while creating a consistent and seamless experience.

Outstanding applications of AI Chatbot

How is AI Chatbot Different from Traditional Chatbots and Virtual Assistants?

Chatbot technology has advanced tremendously since its inception in the 1960s. From interactive FAQ programs based on a limited set of common questions with fixed scripted answers requiring users to choose from simple keywords and phrases to continue the conversation, today we have intelligent AI chatbots with many impressive capabilities such as:

Understanding complex contexts in conversations
Overcoming everything from typing errors to translation issues (Machine Translation)
Mapping meaning with specific intentions that users want chatbots to perform
Creating natural, diverse, and appropriate responses for each user
Communicating in many different languages
Self-improving performance through each interaction without requiring programmer intervention

Comparing AI Chatbot and Intelligent Virtual Agent

Although the initial cost to deploy AI chatbot is higher than traditional chatbots, businesses will save many long-term costs because they don’t need to maintain large customer care teams to handle complex requests or frequently upgrade systems when business needs change.

The latest advancement in Chatbot software is chatbots using Generative AI. They are called “intelligent virtual assistants,” “Virtual Agents,” or “Virtual Assistants,” with famous representatives like Siri, Alexa, Gemini, and ChatGPT.

How does Generative AI improve traditional AI chatbot?

Besides the ability to understand and respond to user questions through Natural Language Processing (NLP) and Machine Learning, Virtual Assistants combine Conversational AI and Robotic Process Automation (RPA) to proactively perform actions on behalf of users.

Comparing AI chatbot and Gen AI Chatbot

For example, when asked about tomorrow’s weather, a Virtual Assistant doesn’t just answer that “it will rain” but also proactively suggests setting an earlier alarm to avoid delays due to rain. Additionally, thanks to Large Language Models (LLMs), Gen AI Chatbot can also:

Create entirely new content (text, images, sound)
Automatically answer questions outside the script scope based on the organization’s knowledge base without pre-programming
Learn from previous conversations to continuously improve how conversation flows are routed
Adapt to each user’s communication style
Express empathy in responses

In other words, AI Chatbot are like helpful friends who know a lot of information, while Virtual Assistants are like actual personal assistants who can act on your behalf.

According to the CEO’s Guide to Generative AI report from the IBM Institute for Business Value (IBV), 85% of executives believe that Gen AI Chatbot will interact directly with customers in the next two years. Enterprise-level chatbots can also integrate into software like Microsoft Teams, creating efficient work centers, and handling from simple tasks to complex processes across different applications, helping businesses automate services and enhance user experiences.

Detailed comparison of the differences between AI Chatbot and Virtual Assistant

In summary, AI Chatbot is intelligent virtual assistant, capable of learning and continuously developing to deliver optimal user experiences. This technology is changing how businesses and customers communicate while optimizing workflows and supporting management. With constant advances in Machine Learning, NLP, and Generative AI, in the future, AI Chatbot will become increasingly sophisticated, bringing convenience and improving service quality for users and businesses.

Text to Speech (TTS) is a technology that converts digital text into natural-sounding audio, allowing computers to read content aloud in a voice that closely resembles human speech. Demand for this technology is growing rapidly, with a projected CAGR of 13.7% between 2024 and 2029. According to Markets and Markets, the TTS market is expected to reach USD 7.6 billion by 2029.

In this article, FPT.AI explores the development, working mechanism, real-world applications, limitations, and future trends of Text-to-Speech technology.

What is Text-to-Speech?

TTS is also known as “speech synthesis” or “computer-generated voice technology”. Most TTS services are offered as APIs, allowing developers to easily integrate voice capabilities into apps, websites, or digital services.

Originally designed to assist individuals with visual impairments or dyslexia, TTS has evolved into a foundational technology powering virtual assistants, automated call centers, and GPS navigation systems. Today, it plays a key role in human-machine interaction and is making the digital world more accessible for everyone.

TTS has become an important foundation for many advanced platforms, enhancing communication between machines and humans

The Evolution of Text-to-Speech Technology

The first electronic speech synthesizer appeared around the 1930s and marked the beginning of TTS development. These early devices had minimal capabilities and were primarily used for research.

How the first electric speech synthesizer worked

In the late 1950s, with the advent of computers, developers began experimenting with algorithms that matched audio files to text components. These early systems produced robotic and unnatural voices.

A major breakthrough came in the 2000s when Deep Learning and neural networks entered the scene. Instead of piecing together pre-recorded sounds, developers began modeling sound waves using real voice recordings.

This shift led to much more realistic, high-quality synthetic voices. At the same time, there were advancements in Automatic Speech Recognition (ASR) and Natural Language Processing (NLP), which laid the groundwork for modern TTS systems.

Milestones in the Development of Text-to-Speech Technology

In the past decade, AI and Machine Learning have further enhanced voice realism—making synthetic speech nearly indistinguishable from human voices. However, this progress also introduces ethical concerns, particularly around audio deepfakes, which mimic real voices without consent. To combat this, tech companies are developing real-time voice detection tools to identify deepfakes and ensure the responsible growth of TTS technology.

The Relationship Between Voice Technology, AI, Deep Learning and Machine Learning

How Does Text-to-Speech Work?

TTS involves both linguistic analysis and speech synthesis. Deep learning models help TTS systems understand how words relate to their audio characteristics and generate realistic AI voices.

Linguistic Analysis

When given a text input, the TTS model first analyzes it using deep neural networks. It examines words, punctuation, and sentence structure to understand intonation, pitch, rhythm, and volume. The system also expands abbreviations, calculates word lengths, determines proper pronunciation, and maps prosody (intonation patterns) across sentences.

How Text-to-Speech technology works

Speech Synthesis

Once the text has been processed, the model converts it into speech using two main steps:

Generate audio features: The model transforms the text into time-aligned features like mel spectrograms, which map changes in sound frequency over time. These features capture detailed characteristics of speech, including pronunciation and emphasis.

The voice is converted by the acoustic model into parameters that simulate the characteristics of human speech

Convert to sound waves: A vocoder model, such as WaveNet or WaveGlow, transforms the spectrogram into an actual audio sound wave that sounds natural. Some TTS systems also allow users to adjust pitch, volume, speed, language, accent, or speaking style

Mel-Spectrogram is fed into the Vocoder, which is converted into an audio signal

TTS systems are built into many devices, such as smartphones, and are available via software, browser extensions, websites, or downloadable apps.

Real-World Applications of Text-to-Speech

TTS is a key component of Conversational AI, especially in applications using Automatic Speech Recognition (ASR) and Natural Language Processing (NLP). It’s a game-changer for people who want to access content hands-free in a fast-paced world.

Here are some major use cases of TTS:

Audio Content: TTS reads digital text, books, lessons, and instructions aloud. News organizations use TTS to convert articles into audio formats for more flexible content access.

Education & Learning: TTS supports students by helping them follow along with text, improve pronunciation, and retain information. It’s especially helpful for people with visual impairments or learning disabilities like dyslexia.

Text-to-Speech technology applications

Virtual Assistants & Chatbots: Virtual assistants like Siri, Google Assistant, and Alexa use both TTS and STT (Speech to Text) to create natural, responsive interactions. They can read messages, make announcements, assist while driving, and offer 24/7 customer support.

How text-to-speech technology integrates into the virtual assistant

GPS Navigation & Maps: TTS enables real-time spoken directions, helping drivers stay focused. It reads street names, traffic alerts, and alternate routes for safer travel.
Multilingual Communication & Language Learning: Apps like Google Translate use TTS to help users understand and pronounce foreign words. It also powers voice-overs for video content in different languages.

Google text-to-speech technology can help users communicate multilingually easily and effectively

Media & Entertainment: TTS creates narration for games, voices for animated characters, and transforms written books into audiobooks, reducing production costs and expanding content accessibility.
Healthcare: TTS reads medical documents, device instructions, and prescriptions to patients. It reminds patients of appointments and medication schedules, especially useful for those with visual or speech impairments.

Application of text-to-speech technology in the healthcare industry

Marketing & Advertising: TTS generates voice content for ads without the need for voice actors. It enhances personalization in campaigns via voice chatbots and email marketing.
IoT & Smart Homes: TTS is built into smart speakers, watches, and home security systems. Devices can speak alerts, schedules, or weather updates, offering seamless, voice-based interaction.

Integration of text-to-speech technology into IoT systems and smarthomes

Customer Service & IVR Systems: TTS powers automated phone systems that answer calls and provide spoken options. When paired with voice recognition, these systems can handle complex queries and deliver voice responses, replacing traditional call center agents.

Text-to-Speech supports providing information and responding to voice commands

Challenges in Implementing Text-to-Speech

Despite its progress, TTS still faces some limitations:

Voice quality still sounds robotic: Some TTS systems still generate flat, machine-like voices that lack natural flow and can hinder listener engagement.

Lack of emotional tone: TTS struggles to convey emotions like happiness, sadness, or surprise, making it less suitable for expressive content like storytelling or film dubbing.

Mispronunciation of special terms: TTS often misreads names, slang, foreign words, or technical terms, leading to confusion in fields like healthcare, finance, or technology.

Incorrect context interpretation: Unlike humans, TTS systems often fail to understand context, which affects rhythm, pauses, and emphasis.

Inconsistent handling of abbreviations: TTS may pronounce the same abbreviation in different ways within a single document.

Incomplete multilingual support: While many TTS systems support multiple languages, they often struggle with mixed-language texts, mispronouncing foreign terms.

Inconsistent tone in long texts: TTS voices can lose consistency across long passages, leading to abrupt changes in tone.

Poor sentence pacing and emphasis: TTS often places pauses and pitch changes in unnatural spots—especially problematic for tonal languages like Vietnamese, Chinese, Korean, or Japanese.

High hardware requirements: Modern AI-based TTS systems require significant computing resources, making them harder to implement on low-power or mobile devices.

Limited voice personalization: While some systems allow basic voice customization, fully cloning or personalizing a unique voice is still a major challenge.

Hạn chế khi triển khai công nghệ Text to Speech

Future Trends of TTS Technology

Here’s what the future holds for TTS:
AI integration to improve voice quality: Advanced AI models like Transformers, WaveNet, and Tacotron are making synthetic voices more human-like. These models can better understand context, adjust tone, and pronounce words accurately across different languages and cultures.
Voice Cloning: This enables TTS to replicate a specific individual’s voice. It’s great for personalized audiobooks, virtual assistants, or customer service bots, making user interaction feel more authentic.
AI Dubbing: This innovation syncs speech with lip movements in videos. It revolutionizes dubbing for films, educational content, and online media by making translations more accurate and lifelike.
Voice Conversion: This allows you to convert one person’s voice into another without re-recording. It’s especially useful in gaming, animation, or podcast production, offering flexible voice creation without additional effort.

Development trends of Text to Speech market according to MarketResearch forecast

In conclusion, Text to Speech technology has become an essential technology in many fields from education, healthcare, marketing to route navigation, virtual assistants, and smart homes. Although there are still some limitations, TTS is constantly improving significantly. The strong growth of the global TTS market reflects the increasingly important role of this technology in improving human-to-machine communication and building a more accessible digital world for everyone.

Digital transformation is spreading strongly around the world, businesses are increasingly facing pressure to optimize operations, AI Agents have become a workforce that promotes innovation activities from internal processes, demonstrating a strategic role in the transformation of modern businesses.

In particular, FPT AI Agents, the platform for creating “AI Human Resources” of FPT Corporation, has been opening a new era for domestic Vietnamese enterprises with outstanding potential in integrating, operating and developing internal automation solutions, meeting domestic market needs, while competing internationally.

Innovating internal business operations by implementing AI Agents for business

Businesses around the world have quickly recognized the importance of internal automation to minimize repetitive tasks, minimize errors and optimize resources. AI Agents systems are now widely applied in process automation, helping businesses handle work quickly and accurately. According to reports from McKinsey and Gartner, the application of automation technologies can help businesses reduce internal process processing time by 50% to 80%, while cutting personnel costs by 30% to 50%. Pioneering applications such as JPMorgan Chase’s internal system (reducing 360,000 labor hours annually) or IBM Watson integrated into the internal management system have demonstrated the outstanding effectiveness of AI in improving productivity and business efficiency. These achievements not only open up great cost savings opportunities but also facilitate businesses to make decisions based on accurate and fast data.

In the global context, 2025 marks the time when businesses will be ready to comprehensively transform with the support of AI Agents. In particular, the Vietnamese market – with its strong digital transformation policy, investment in IT infrastructure and human resource development – is becoming a potential place to apply these automation solutions, helping domestic businesses compete in the international arena. In May 2024, FPT became a strategic partner with NVIDIA through a $200 million investment in the development of an AI Factory. At the Tech Day 2024 event, FPT officially launched FPT AI Factory – a set of solutions to support the development of the entire AI process. In this ecosystem, FPT AI Agents is positioned as the core platform that allows the construction and operation of “AI personnel” quickly and flexibly. FPT AI Agents not only “Vietnamizes” the interface and content to suit the local culture but is also built on an advanced technology platform, allowing businesses to quickly deploy and customize according to business requirements. As a result, businesses in Vietnam can achieve flexibility, improve service quality and increase productivity, opening up opportunities to reach the international market.

FPT AI Agents is developed by FPT.AI based on advanced Generative AI technology and large language models (LLM) combined with Deep Learning and Reinforcement Learning. This platform allows businesses to easily create a multilingual AI team (including Vietnamese, English, Indonesian and Japanese) without requiring in-depth programming or technical knowledge. Users only need to interact in natural language to control and communicate with the system, thereby empowering human employees, revolutionizing customer experience and breakthrough business productivity. These technologies help FPT AI Agents not only perform tasks accurately but also automatically learn, update new knowledge, create contextually appropriate responses and remember previous conversations.

>>> EXPLORE: How to build an AI Agent and train it successfully?

Applying FPT AI Agents in internal operations

AI Agents – Optimizing Administrative – Human Resources Operations

The administrative department plays a fundamental role in the operations of every business. From managing records, processing text documents, planning and scheduling, supporting departments to ensure smooth business operations. However, with a huge amount of work, the dependence on manual processes makes administrative work often time-consuming, error-prone and unable to keep up with the pace of modern business development.

The emergence of AI Agents – artificial intelligence assistants – has opened a new turning point in administrative management. In particular, FPT AI Agents can automate repetitive tasks, improve work productivity, reduce administrative workload and optimize operating costs. According to McKinsey’s report, the application of AI Agents in the administrative department reduces 50-70% of manual workload, increases employee productivity by 40%, cuts up to 30% of administrative operating costs, and helps employees focus on strategic and creative tasks.

AI Agents are capable of handling key operational and human resources tasks professionally:

Document management and automatic text processing: AI Agents can scan and extract information from documents (PDF, Word, images); automatically classify and store documents for easy searching; convert paper documents into digital data with OCR (Optical Character Recognition), reduce document processing time by about 60%, increasing data entry accuracy by 90%. In particular, AI Agents can manage personnel records, helping to track labor contracts, and periodically update social insurance.
Manage work schedules, meetings and meeting rooms: AI Agents automatically check available schedules and schedule meetings that suit everyone; Send invitations, schedule reminders via email, chatbot or text message; Record meeting minutes, summarize main content. This can help reduce 50% of the time to organize meetings, limit scheduling conflicts between departments
Optimize reporting and data analysis processes: AI Agents automatically aggregate data from multiple sources, analyze, create reports and suggest improvements, display visual data on dashboards, helping to reduce about 80% of reporting time, increase 45% accuracy in data analysis.
Smart recruitment process: automatically screen thousands of profiles based on pre-defined criteria, helping to reduce selection time and increase the ability to find suitable candidates, creating an effective recruitment process.
Internal human resource training: AI Agents have the ability to support businesses in improving the quality of human resources by automating and personalizing the training process through the application of modern learning methods.

FPT AI Agents

In Vietnam, FPT AI Mentor has become a powerful assistant accompanying all employees, answering all questions related to work, helping employees improve their knowledge and abilities, thereby increasing productivity and operational efficiency for the business. FPT AI Mentor will assess employees’ daily knowledge, personalize training content according to each individual’s strengths and weaknesses. Knowledge in the following days will focus on weak areas, combined with the spaced repetition method to maintain solid knowledge. In addition, employees can access knowledge anytime, anywhere thanks to diverse learning content formats such as text, images, audio and short videos, built on corporate knowledge sources. At the same time, FPT AI Mentor provides a professional reporting system for all levels. For employees, FPT AI Mentor will build a knowledge map based on the assessment results, helping employees understand their existing and needed knowledge and propose the most suitable learning path. Managers can quickly grasp the training quality of employees with summary reports by department and campaign, customized according to the needs of the business.

Many businesses applying FPT AI Agents in human resources have recorded a 50% reduction in recruitment time and an increase in productivity of up to 40% thanks to the ability to automate and personalize the training process. Banks and financial institutions have used FPT AI Agents to automate the process of processing invoices, preparing financial reports and reconciling transactions. This helps reduce errors and improve the speed of decision making based on real-time financial reports. Integrating FPT AI Agents into CRM and ERP systems has improved customer information processing, optimized data management and enhanced customer experience through the automation of multi-channel customer care tasks.

>>> EXPLORE: Why Gen AI Agents are the future prospect of Generative AI?

AI Agents in the Finance and Accounting Department: Transparent, accurate and fast

The finance and accounting department plays an important role in the operation and sustainable development of the enterprise. However, the large workload, high accuracy requirements along with risks of fraud, errors and cash flow management require businesses to continuously improve to optimize the finance and accounting process. Accounting operations require high accuracy with a large workload. According to PwC’s Report, AI Agents can help businesses reduce 80% of manual work in accounting, increase accuracy up to 99%, and save 30-50% of operating costs.

AI Agents are capable of processing invoices and documents by automatically scanning, extracting and storing information from invoices, contracts and financial documents, helping to reduce errors and increase processing speed by up to 80%. In particular, the system automatically synthesizes data from multiple sources, creates financial reports, real-time revenue and expenditure, supporting the management in decision making. In addition, AI Agents use analytical algorithms to compare transactions with standard samples, provide early warnings of abnormalities and risks, and detect fraud quickly and effectively. Application in the accounting department not only helps reduce operating costs but also increases data accuracy, thereby supporting businesses in making decisions based on updated and accurate information.

Automate data entry and invoice processing: AI Agents help automatically extract information from invoices & contracts using OCR (Optical Character Recognition) technology; Assign tax codes, automatically account for and check for duplication; Synchronize with accounting & ERP systems for faster processing. This helps reduce accounting data entry time by about 70%, increase accuracy in invoice processing by 95%.
Financial data analysis and trend forecasting: AI Agents help analyze financial data in real time, predict revenue, costs and market trends to help businesses make strategic decisions and assess financial risks, optimize budgets and allocate capital more effectively. As a result, AI Agents support a 40% increase in accurate cash flow forecasting and a 30% reduction in financial risks thanks to accurate data analysis.
Fraud Detection and Risk Control: AI Agents help monitor unusual transactions, alert fraud immediately, compare transaction data with financial history to detect discrepancies, ensure compliance with financial standards, and minimize the risk of legal violations.
Support for automatic financial reporting: AI Agents help automatically synthesize data from ERP & CRM systems, create financial reports according to international standards, help businesses make more accurate decisions and provide intuitive dashboards, helping managers easily monitor finances.

Applying AI Agents to finance – accounting not only helps businesses reduce manual work but also creates transparency, accuracy and speed in operations. With the support of FPT AI Agents, Vietnamese businesses can automate data entry, optimize cash flow, detect fraud and improve the efficiency of financial analysis.

>>> EXPLORE: Understanding AI Agents in KYC

AI Agents in Marketing and Sales: Winning Customers and Boosting Revenue in the Digital Age

In the modern business context, where customers increasingly expect personalized experiences and responsiveness, the sales & marketing department plays a central role in driving revenue and building sustainable customer relationships. However, businesses are still facing many challenges in optimizing internal processes, synchronizing data and improving work performance. These issues not only reduce sales efficiency but also waste resources, increase costs and reduce customer satisfaction.

In response to that need, FPT AI Agents – FPT’s advanced “AI human resources” creation platform – appears as a comprehensive solution, helping to automate sales and marketing processes, increase productivity and improve customer experience from within. FPT AI Agents has outstanding strengths such as multilingual support (Vietnamese, English, Japanese, Indonesian – suitable for domestic and international businesses), quick deployment (building AI Agents in just a few hours, easy integration with CRM, ERP systems), analyzing customer data to create interactions suitable for each subject, helping to personalize customers optimally. In particular, FPT AI Agents ensures information security issues, complies with international standards on data security. With support from the 200 million USD investment with NVIDIA and the FPT AI Factory platform, FPT AI Agents owns a powerful “super infrastructure”, allowing businesses to comprehensively improve sales & marketing performance.

FPT AI Agents is applied to internal sales and marketing operations such as:

Managing customer data and optimizing CRM systems: Customer management is the core foundation in all sales & marketing activities. However, when data is scattered and not updated promptly, businesses easily lose opportunities to close deals and waste time on manual processing. FPT AI Agents helps automatically import and update data from the AI system that automatically collects information from channels (email, social networks, websites) and updates it into CRM; intelligently classify customers based on behavior, purchase frequency and interaction, AI Agent groups customers to prioritize approach; track customer journey, support sales teams to grasp the status of each customer to develop appropriate strategies.
Automate sales processes: Traditional sales processes often take a lot of time to schedule appointments, create quotes and track orders. FPT AI Agents helps schedule appointments and automatically remind, ensuring that the sales team does not miss opportunities to approach customers; create quotes and contracts quickly with the AI system that supports creating and sending quotes in a short time; Personalize consulting with product/service suggestions that suit each customer’s needs.
Support internal marketing campaigns: For a marketing campaign to be highly effective, there needs to be smooth coordination between the marketing and sales teams. FPT AI Agents supports automatic sending of emails and promotional messages with personalized content based on customer behavior; monitor campaign performance with AI Agents’ analysis of email open rates, clicks and conversions, helping the team adjust strategies in a timely manner. At the same time, AI Agents can make consumption trend forecasts from data analysis via social networks and customer behavior to make accurate predictions.
Train and support new sales staff: Training new staff is time-consuming and costly. FPT AI Agents helps provide documents and automatic instructions, new staff can easily look up sales processes and product information; support direct situation handling, answer questions about products, sales policies immediately and make suggestions for development roadmaps based on work performance, AI suggests appropriate training.

FPT AI Agents not only automates processes but also brings a series of practical benefits, saving 50% of work processing time, increasing 35-40% conversion rate thanks to accurate data and quick response, shortening order closing time, helping revenue grow significantly. Thereby creating innovations in customer experience, supporting customers to make quick decisions.

2025 promises to be an important milestone marking the explosion of AI Agents solutions in businesses. With the global trend and strong development of AI technology, businesses will continue to digitally transform, integrate automation systems to optimize internal operations, thereby enhancing business efficiency and competitiveness. Applying AI Agents in internal operations is not only an inevitable trend of digital transformation but also a breakthrough that helps businesses optimize operations, minimize risks and costs, and improve productivity and employee experience. With FPT AI Agents, businesses in Vietnam can create a modern, flexible and creative working environment, thereby strengthening their competitive position in the domestic and international markets, opening up a sustainable development future in the new era.

>>> EXPLORE:

AI Agents are becoming the mainstream trend in 2025 with the ability to support and replace humans in performing many tasks, from simple to complex. According to many experts, the explosion of AI Agents in work is a “stepping stone” before humanity moves towards artificial general intelligence AGI.

Mr. Nguyen Hong Phuc, Chief Science Officer of Conductify AI, commented that artificial intelligence has a much faster and more comprehensive development speed than other technologies, such as biology or quantum computing. “On average, AI has a small breakthrough every two weeks, a big step forward every month, and in the past year, AI has made comprehensive changes,” he shared.

2024 will witness a strong expansion of AI, from the ability to generate text like ChatGPT in 2023, to new areas such as image, video, audio creation, content creation and learning support. However, the question is: How effectively can AI change the way people work?

Most AI models, such as OpenAI’s GPT and o, Google Gemini, Anthropic Claude, Meta Llama or DeepSeek, mainly help speed up productivity rather than completely change the way people work. People still do the same work as before, but at a faster speed and higher efficiency. The birth of AI Agents marks an important turning point, when AI not only supports but also has the ability to automate processes and perform work on behalf of humans more proactively.

In this article, FPT.AI will describe in detail the trend of applying AI Agents at work and why AI human resources are a prominent technology trend in 2025. Let’s explore.

What are AI Agents?

At the end of 2024, the term AI Agents began to be mentioned a lot by large technology corporations and scientists around the world. In fact, this concept has been around since 2023, referring to the ability of artificial intelligence to assist with repetitive and boring tasks.

“Imagine an AI in customer service that can predict a user’s needs before providing an answer, or an AI managing network connections that can detect potential problems and automatically handle them to ensure uninterrupted service,” Liz Centoni, Cisco’s Director of Customer Experience, shared with National Technology. “This will be an important trend in 2025, when AI becomes more autonomous, more automated, and integrated into personal devices, IoT, robots… to perform tasks in reality.”

According to Mr. Phuc, AI Agent can be understood as an artificial intelligence system that is capable of operating independently and making decisions based on specific goals set by humans.

“Unlike AI chatbots or virtual assistants that can only answer questions, AI Agents are designed to handle complex tasks, can interact with other systems and even operate continuously without human intervention. In other words, AI Agents at work not only respond but also proactively complete tasks and make decisions when necessary,” Mr. Phuc explained.

According to Mr. Tran The Trung, Director of FPT Technology Research Institute, in general, AI Agent is an automatic system that performs professional tasks in a similar way to humans or employees in the organization, but without support or intervention and still achieves high efficiency.

According to IBM’s definition, AI Agent is a system or software that can automatically perform many tasks on behalf of humans, by building the optimal operating order and taking advantage of available tools.

Amazon has also provided a similar definition, describing an AI Agent at work as a software program that can interact with its environment, collect data, and use that information to perform self-defined tasks to achieve a given goal. While humans set the goal, an autonomous AI Agent can make its own decisions about what actions to take to achieve that goal.

AI Agent is a system that can automatically perform many tasks instead of humans

>>> EXPLORE: How to build an AI Agent and train it successfully?

Real-life applications of AI Agents work

According to Mr. Phuc, the application of AI Agent in work brings a breakthrough when compared to chatbots. Instead of just stopping at the question-and-answer function, AI personnel can automate work processes flexibly.

“We are getting closer to AI replacing humans in repetitive tasks, and that can start as early as 2025,” he commented.

Technology expert Duy Luan assessed that AI Agent, to some extent, can think for itself and use designated support tools. “This technology can be applied in many fields, typically controlling computers, retrieving and processing information according to input, supporting customers,…” Mr. Luan shared.

In the world, a number of large technology corporations have been considering using AI Agent in work to replace humans. In January 2025, Meta CEO Mark Zuckerberg said that AI Agents could take over the role of a mid-level engineer in his company as early as this year.

In December 2024, Salesforce, one of the leading cloud technology companies, announced that it would not hire any more software engineers in 2025, citing the significant improvements in productivity thanks to AI. The company has developed an automated AI Agent with flexible customization capabilities, directly connected to business data, and performing various tasks in the fields of sales, customer service, marketing, and commerce.

Meanwhile, OpenAI CEO Sam Altman said that virtual employees will start entering the labor market this year. In a blog post on January 6, he shared:

“2025 will be the year we witness the emergence of AI Agents, also known as virtual employees, in many businesses. They not only support but can also significantly improve work productivity. The gradual introduction of advanced AI tools into practice will bring about large and increasingly widespread impacts.”

The Huge Potential of the AI Agents Market

In a report predicting 6 AI trends that will explode in 2025, Microsoft Vietnam assessed that AI Agents will reshape the way we work: “Thanks to advances in memory, reasoning and multimodal interaction, AI Agents can handle complex tasks more flexibly and effectively.”

Sharing the same view, Mr. Nguyen Nhu Dung, Managing Director of Cisco Vietnam, Laos and Cambodia, also predicted that AI will become an important part of the workforce, taking on many tasks instead of humans.

>>> EXPLORE: AI Agents for Business Internal Operations in Vietnam

In Vietnam, how is the trend of applying AI Agents at work taking place?

“Vietnam will have thousands of AI personnel, or in other words, thousands of mature AI models, widely applied in all areas of life. These assistants can support millions of people at the same time, helping to increase labor productivity many times over. When AI Agents are fully automated, AI Agents will perform work at a speed far surpassing that of humans” Mr. Truong Gia Binh shared about the future development of artificial intelligence in Vietnam in an interview with VnExpress in December 2024.

Two months earlier, FPT launched its first AI Agent at the Techday 2024 event. According to Mr. Vu Anh Tu, FPT’s Chief Technology Officer, FPT AI Agents is a platform researched and developed by the group’s experts to create and operate multilingual AI Agents, leveraging generative artificial intelligence (Generative AI) and large language models (LLM). This platform helps businesses build a team of AI personnel capable of collaborating effectively with humans.

Mr. Tu said: “AI personnel will appear everywhere, supporting people in many different tasks such as programming, processing documents or writing emails. We aim for every FPT employee to have an AI Agent at work to work more effectively, and each customer will also have at least one AI Agent to support them. This year, we will accelerate this process.”

Applications of FPT AI Agents in businesses

According to Mr. Tran The Trung, Vietnam already has a strong computing infrastructure, creating favorable conditions for developing artificial intelligence algorithms to serve the development of AI Agents. He also emphasized that “Some units have collected and accumulated large amounts of data to train AI Agent models”.

In addition, he revealed that some domestic companies are implementing humanoid robot projects to combine with AI Agents. “Vietnam has the opportunity to catch up with the growing AI Agent trend in the world. However, this will depend on the level of awareness, action strategies of pioneering enterprises as well as support from the Government,” Mr. Trung commented.

In short, the appearance of AI Agents in work is creating important changes, not only helping to speed up work but also gradually automating complex processes. Major technology corporations in the world and Vietnam have been investing heavily in building infrastructure and developing AI Agent models, to seize the opportunity to optimize labor productivity.

Although there are still challenges in deployment and management, the potential of AI Agent in work is undeniable, bringing us closer to a completely new era of labor.

>>> EXPLORE: