Hanoi Forum Launches ‘Agent City’ Initiative to Let AI Agents Work Across Public Services
An initiative to build a large-scale environment where AI agents carry out tasks across public and commercial services under shared rules was launched in Hanoi on October 10, Kenh14 reported. The Agent City initiative, run by Tyr, a company building security infrastructure for the AI agent economy, together with the Stamford Global Leadership Center (SGLC) and the Institute of Innovation and Digital Transformation (VIDTI), was unveiled at the Smart Agent City forum, part of events marking National Digital Transformation Day 2026.
Hoang Huu Hanh, deputy director of the National Digital Transformation Agency, described three stages of urban development: the digital city, the smart city and the city run by AI agents. Today’s systems can spot traffic jams and flooded streets, he said, but cities need help acting on that information, for example by rerouting vehicles, warning residents and organising rescues during floods. For agents to do that, data must be standardised, machine-readable and shared, while privacy and security are protected.
Bangdao Chen, co-founder of Tyr and of the Blockchain Research Centre at the University of Oxford, said the initiative has three priorities: data sovereignty and autonomy, security in how agents operate, and narrowing the digital divide. Its guiding principle is that AI does not govern people; people manage and control AI. Each person or organisation keeps control over its data, internal rules and the agents acting for it, and an agent can be stopped before it takes an action beyond its permissions. To reach another organisation’s systems, an agent must pass a checkpoint that verifies its identity, confirms whom it represents and grants only the access the task needs.
What it means in practice
Panellists stressed that the model is at an early stage. Chen urged Vietnam to start with testing and simulation to spot risks and refine governance before any wide rollout, and to reach residents through familiar channels such as email and messaging apps rather than new accounts. Timothy Ngo of Iozera.ai said agents depend on data centres, power, water and skilled local engineers, and on community support for that infrastructure. Ashish Arvind Thukral of NeuGenM.ai said businesses will need product data that agents can find and understand, plus safe rules for agent-to-agent transactions. Reports from the forum did not mention a budget, pilot site or timeline.
The idea extends a shift already under way among Vietnamese companies, from chatbots toward autonomous AI agents. Its emphasis on human sign-off also echoes the State Bank of Vietnam’s risk tiers and human review rules for banking AI.
What to watch: whether Hanoi or another city offers a sandbox for the first agent services, which public services are chosen, and how the checkpoint rules fit Vietnam’s personal data protection requirements.