In step with every beat of the flow of money

For more than twenty years I have studied, researched and worked in artificial intelligence, from the days of writing C code to read individual pixel values, to today, when AI can read and understand an entire credit file. During that time I served as Director of the AI Center at a major Vietnamese bank. In that role I found new ways for AI to deliver lasting value to a bank, and led its end-to-end AI transformation, from strategy, data foundations and teams to applications running on real operational systems. It taught me lessons that no textbook records.

The first lesson, which I carry with me to this day, is that money has a rhythm of its own. It changes constantly and never rests, driving the growth of the economy and the prosperity of society. A transfer in the middle of the night, or an unusual transaction at two in the morning, may point to fraud. A file is disbursed late because of a shortage of staff, a public holiday, or an approval process that has not yet been digitised or fully brought into the age of AI. A customer's question on a weekend night goes unanswered. Each of these beats needs to be seen, understood and handled in time. People have their limits; they need rest and a life to enjoy. Money, however, must keep moving to keep the economy moving. AI Agents are the tool every bank must have so that people can keep pace with the flow of money and push it forward.

Two decades with AI have let me watch this technology pass through many turning points: from models that recognise and predict objects and events, to models that understand language and can hold a conversation like a person. But I believe the most important stretch has only just begun: AI moving from answering questions to completing the work it is given. That is Agentic AI, where AI Agents plan, have skills to handle tasks, use tools, work with one another, and take responsibility for every step they carry out.

My years inside a bank taught me something else: to follow and move the flow of money, an AI Agent must stand where the money flows. That means the core banking system, credit appraisal, anti-money-laundering, CRM and data warehouses, the systems a bank has spent decades building and running. However intelligent an AI Agent is, if it stays outside those systems it only watches the flow from a distance and cannot move it. MirAI chooses to focus on the hardest part of the job: integrating Agentic AI applications into the bank's existing IT infrastructure, respecting the architecture already in place rather than demanding that it be replaced. Once properly connected, always-on agents can stay close to every transaction, every file and every conversation, day and night, in the bank's real operations.

For us, a full AI transformation is more than a handful of scattered use cases. It is a journey toward AI being present at every stage of the business flow, in which a bank needs to (1) redefine which processes should be handed to AI Agents, and (2) prepare people, data and governance so that Agents and staff can work side by side as colleagues. The Financial AI Factory Platform (FAFP) was created for that journey: a "factory" where a bank designs, tests, deploys and monitors AI Agents to one consistent standard. Together with Finance Advisor (FinAD), the Atobar.ai credit scoring solution, AI-FP, AI-CFO, AI-CR, AI-FM, AI-CX, specialised Agentic AI solutions and voice recognition solutions, we hope to stand beside you from the very first problem to the day AI becomes a natural part of everyday operations.

To follow the flow of money is also to carry responsibility for it. Finance is built on trust, and that trust does not tolerate risk or error. So before it goes into operation, every AI Agent must be able to answer three questions:

1. Is the Agent safe?

   How is customer data protected, how far is access limited, and are important decisions approved by a human?

2. Is the Agent compliant?

   Is every action logged for audit, and does it fit the rules of regulators in Japan and Vietnam as well as the bank's own internal policies?

3. Is the Agent worth what it costs?

   Models must be chosen to suit each task and running costs tracked down to each Agent, so that the value created always exceeds the cost incurred. An AI Agent that costs more than the value it brings becomes a burden, however intelligent it is.

The future does not arrive by itself; it is built every day by specific people. MirAI's engineers are young talents in AI, many of whom grew up inside Vietnam's leading banks. They understand what a bank needs from AI, and how AI is applied inside a bank. Together with MILIZE in Japan, we connect Vietnam's AI technology capability with the standards and depth of expertise of Japan's financial market.

We do not claim to be ahead. What we can promise is the care and diligence of experienced people: listening closely before we propose, testing carefully before we deploy, and staying alongside you long after handover.

My wish is a simple one. With every beat of the flow of money, day or night, there will be AI Agents built by MirAI together with you, quietly keeping pace: safe, compliant, and creating lasting value for the bank with every beat.