Lan also predicts that banking teams could shrink to three core employees who are supported by roughly ten AI agents. This would mean far fewer staff are needed for routine work.

According to Nguyen Anh Tuan, Techcombank’s adoption of AI and Copilot tools has already lifted employee productivity by 25‑30 %. The bank serves 18 million customers with about 12 000 staff, showing that AI can boost output without a large increase in headcount.

Vu Thanh Tung, CEO of GreenNode, warns that the hardware required for large‑scale AI is very expensive. A server equipped with eight GPUs costs roughly VND20 billion (about $760,000). He also notes that Vietnam lacks enough professionals who can operate such infrastructure, creating another barrier for banks.

Senior Lieutenant Colonel Trieu Manh Tung of the Ministry of Public Security says criminals are already using AI to create fraudulent payment messages and deep‑fake videos that can bypass biometric checks. This raises serious cybersecurity worries for banks and their customers.

In the coming three years AI agents are likely to handle half of banking transactions, cut staff numbers, raise productivity, but also demand costly hardware, skilled workers, and stronger security. Understanding these shifts will help customers and job‑seekers prepare for a new banking landscape.