People first, then technology: Lessons from cities leading on AI
Cities worldwide are deploying AI to improve public services, prioritizing human needs and data control, with measurable gains in efficiency and satisfaction.
Jakarta uses AI to predict flooding up to six hours in advance, enabling authorities to close gates, activate pumps, and warn residents before water levels rise. The system relies on trusted data to provide actionable forecasts that reduce response times during emergencies. Aberdeen City Council’s AI-powered virtual assistant now delivers specific answers to residents 24/7, with plans to expand language support based on local needs. The upgrade was developed in collaboration with Microsoft to ground responses in verified council information.
Raleigh, North Carolina, partnered with Microsoft, NVIDIA, and Esri to create an AI-driven digital twin of its traffic system in six weeks. The city estimates that reducing delays by 4.5 seconds at 10 key intersections could save commuters $9.7 million annually in time and travel costs. The project highlights how AI can optimize infrastructure without lengthy development cycles. ThoughtLab’s research surveyed 525 officials and 50 urban leaders across 200+ cities to assess AI adoption in urban governance.
Abu Dhabi’s TAMM platform, powered by Microsoft, provides over 1,200 government services to 4.7 million users. Customer satisfaction rose from 78% to 94.5% after AI integration, while service costs dropped by 30%. TAMM AutoGov uses agentic AI to proactively renew trade licenses and disability cards, reducing resident-initiated requests. The platform was recognized as the world’s best e-Government project in 2025 by the UN-backed WSIS Prizes.
Finland’s Tampere VIP assistant, built by two local developers and Microsoft, uses multiple AI agents to answer resident queries about dining, events, and activities. The platform can be reconfigured for new purposes by swapping agents and data sources without rebuilding. Munich Fire Department’s AI voice operator, now in testing, handles non-urgent patient transport bookings, freeing dispatchers to focus on emergencies. The system escalates uncertain calls to human operators, ensuring accountability in high-stakes decisions.