Cognitive Cities: How AI Can Help Cities Move From Seeing Conditions to Acting on Them

Picture 60,000 fans leaving a stadium. Cameras can see a crowd building at one exit, but seeing it does not clear it. Someone must decide whether to open another gate, redirect fans or shift staff, and do it in time. Intersections, hospitals and stores face the same challenge: people are moving, conditions constantly change, safety matters and operations cannot stop.

For years, a "smart" city or space was largely about connectivity. Sensors detected what was happening, cameras provided visibility and dashboards delivered more information. AI creates an opportunity to go further by helping places understand context, determine what should happen next and turn intelligence into action. That is the shift from a smart city to a cognitive city.

Cognitive systems connect insight to action

I think about this evolution through a simple loop: Sense. Understand. Decide. Act. Learn.

A connected system can tell an operator that traffic is backing up. A cognitive system can put those conditions in context and help determine the appropriate response. The same principle applies when a utility responds to a pressure drop, a hospital coordinates resources or a stadium manages changing crowd flows.

Intel sees three complementary layers enabling this shift:

• Edge AI processes data where it is created.

• Agentic AI turns intelligence into decisions.

• Physical AI carries those decisions into machines and infrastructure.

This is not about removing people from the equation. It is about helping people operate increasingly complex environments, with human oversight where it matters.

Cities are beginning the transition

Most places haven't made that shift yet. According to ThoughtLab's Building a People-First, AI-Powered City study, only 17% of cities qualify as AI Leaders. The study reports that these Leaders score 64 on advanced AI capabilities, including agentic AI, physical AI and digital twins, compared with 37 for other cities.

The study also projects that AI Leaders' use of agentic AI will rise from 42% to 84% in three years, while their use of physical AI will increase from 47% to 79%. Together, these findings point to a growing gap between cities beginning to connect intelligence with action and those still building foundational capabilities.

Where that intelligence runs also matters. In Sarasota, Florida, computer vision using existing traffic cameras was connected to the traffic-management system. Intel and Derq report a 33% reduction in crashes, including a 90% reduction at one intersection. This is one deployment, not a guarantee of results elsewhere, but it illustrates the pattern: local detection connected to an existing system in time to support action, while the cloud continues to train models and analyze information across sites.

Build on the systems cities already have

The barriers to broader adoption are real. ThoughtLab reports that 84% of cities worry about unexpected costs, 82% lack AI skills, 81% cannot prove the return on investment and 76% are constrained by siloed data and overlapping point solutions.

Read that list as a product specification: lower total cost, solutions that integrate rather than create another silo, tools existing staff can use and a clear path to demonstrating return. Cities have already invested in cameras, sensors and operational systems. The next step should build on those investments, not automatically replace them.

Create a shared foundation, then innovate on top of it

Cities cannot treat every AI opportunity as a new technology stack. Mobility, safety, utilities, healthcare, sustainability and public services have different missions, but they increasingly require the same core capabilities: local intelligence, trusted data, security and governance, and infrastructure that connects with other systems. If every use case produces another technology island, AI risks making fragmentation worse.

That is the idea behind Intel's Cognitive Cities approach: a common Edge AI foundation designed to be:

• Open, to preserve choice.

• Local, to support governance requirements.

• Efficient, to manage power and cost.

• Proven, with a path from pilots to deployment.

Intel provides a platform spanning silicon, systems and software, including Open Edge Platform and Metro Ai Suite that enables cities to layer edge AI, agentic AI and physical AI capabilities, while partners build services on top. Intel's role is not to build the city. It is to provide a foundation on which others can build and differentiate. Intel's expansive, open ecosystem approach gives cities a lot of vendor choice to quickly achieve their objective.

Keep people at the center

As AI becomes more consequential in the physical world, trust becomes more important, not less. ThoughtLab reports that 78% of cities cite over-surveillance and loss of privacy as a barrier. The study also finds that 79% of AI Leaders have formal transparency guidelines, compared with 34% of other cities. It reports that 68% of Leaders limit the data they collect and retain, compared with 49% of other cities.

These findings make guardrails product features, not add-ons. Cognitive systems need clear authority levels, audit trails, local control and a human in the loop when the stakes are high.

A cognitive city should not act simply because technology makes action possible. It should help people make better decisions, automate appropriate tasks and preserve human judgment where it matters most.

Start with one place and one meaningful problem

Cities do not need to wait for a fully autonomous future. Start with one space, one use case and one partner, and build on a foundation that allows the next use case to move forward without starting over.

The promise of a cognitive city is not a city that runs itself. It is a city that helps people see sooner, decide better and act when it matters.

To learn more visitintel.com/cognitivecities.