In a significant move towards modernizing its software development lifecycle, Techcombank has announced the launch of a modular AI agent platform that drastically reduces the time needed to develop functional agents. Previously, building an agent with full functionality took around eight months; now, the bank claims it can accomplish this in less than a week.
This initiative is part of Techcombank's broader digital transformation strategy, which has involved substantial investments of approximately $1 billion in recent years. Serving over 13 million retail e-banking customers, the bank has reported impressive growth metrics, including a 21% increase in assets and an 18% rise in net profit for 2025. The focus on enhancing productivity through AI is evident, with the cost-to-income ratio improving from 32.7% to 30.8% during the same period.
Centralised AI Governance
Central to Techcombank's strategy is the establishment of a centralized AI governance framework. Initially, various teams attempted to implement AI solutions independently, leading to potential duplication of efforts and fragmented investments. To address these risks, the bank has instituted a structured approval process for AI initiatives, overseen by its chief executive officer. This governance structure ensures effective coordination of AI projects across engineering, retail, corporate banking, operations, human resources, and security.
Tran Hoang Quan, the bank's chief technology officer, emphasized the importance of security in the AI deployment process. "All AI models, training data and outputs are hosted internally within Techcombank's environment. Nothing goes outside Techcombank," he stated. This commitment to secure data management influenced the decision to utilize Amazon Q, as it met the bank's stringent internal hosting and data protection standards.
The Modular AI Agent Platform
The newly launched agent platform, called the 'Virtual Squad', is designed to facilitate the development of AI agents capable of supporting various roles within the software development lifecycle. Built on open-source frameworks such as LangChain and LangFlow, the platform incorporates retrieval-augmented generation (RAG) architecture and utilizes Amazon Bedrock for hosting foundation models.
Tran highlighted the platform's versatility, noting that it is designed for reuse across different roles. For instance, the business analyst capability includes multiple sub-agents that can conduct requirement analysis, impact assessment, and use case generation, allowing these functions to be shared across various teams.
Implications for Productivity
The primary objective behind this new platform is to enhance productivity in software development. As Tran noted, key drivers include the reduction of design, coding, and testing durations through AI agents tailored for roles such as business analysts, quality engineers, enterprise architects, solution architects, IT project managers, and developers.
With this strategic shift, Techcombank aims to accelerate its internal software development processes while setting a precedent for how financial institutions can effectively utilize AI to streamline operations and reduce costs. As AI adoption becomes increasingly critical in the banking sector, Techcombank's approach may serve as a model for other financial institutions looking to capitalize on the potential of AI in their operations.
The bank’s commitment to a secure, centralized governance framework, combined with its innovative agent platform, positions it well for future advancements in AI and digital transformation. As the technology landscape continues to evolve, Techcombank's proactive measures could lead to more efficient practices and increased competitiveness in the financial sector.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
