From the Factory Floor to the City Street
AI is no longer confined to data centers and lab environments - it is quietly, but rapidly, reshaping the physical spaces where millions of people live, work, and move every single day.
At the IoT Tech Expo North America, a compelling fireside chat brought together Sandra Baer, CEO of Personal Cities, and Jumbi Edulbehram, Director of Global Public Sector at NVIDIA, to explore how artificial intelligence is becoming the connective tissue of tomorrow's cities. The conversation surfaced two transformative forces- Agentic AI and Physical AI and asked a deeper, harder question: are our cities ready to collaborate enough to make this work?
01 Agentic AI: Cities That Think
Agentic AI refers to the emerging generation of large language model (LLM)-powered systems that don't simply answer questions, they reason, plan, and act. Unlike static software, these agents can be trained on city-specific data, retrieve real-time information, and communicate findings in plain language that residents, planners, and policymakers can actually use.
The potential here is enormous. Consider the housing crisis: a well-trained urban agent could simultaneously model traffic impact from a new development, forecast strain on public transit, simulate construction timelines, and flag zoning conflicts- all in minutes, not months. These are exactly the kinds of cross-domain problems that have historically paralyzed city governments, not because the data doesn't exist, but because no single team can process it fast enough.
"Agents can do reasoning with AI like a person and we can train them on a regular basis to keep pace with a changing city."
Beyond planning, agentic systems open doors to revenue-generating city applications: streamlined permitting workflows, predictive maintenance alerts for aging infrastructure, and automated service dispatch that cuts response times and costs simultaneously.
02 Physical AI: The City That Sees
If Agentic AI is the brain, Physical AI is the nervous system, the layer of sensors, cameras, and edge devices that connects digital intelligence to the real world. The speakers distilled Physical AI into three interconnected pillars:
Perception
Cameras and sensors that observe the physical environment — traffic density, crowd flow, infrastructure stress — in real time.
Simulation
AI trained on historical patterns to model future scenarios — such as predicting gridlock before it occurs and rerouting dynamically.
Generation
Turning physical-world data into concrete actions — dispatching emergency services, adjusting signal timings, issuing alerts.
03 The Collaboration Crisis
Perhaps the most candid and important part of the session was the acknowledgment that technology is not the primary bottleneck. Collaboration is.
Cities are islands of data. For example, a traffic management system in one city operates largely in isolation — even though congestion from neighboring cities flows directly into its streets. The problem compounds at regional scale: emergency services, transit networks, utility grids, and environmental monitoring systems all generate valuable data, but share almost none of it across municipal boundaries.
The barriers are partly technical, but mostly organizational and political: differing data standards, legitimate privacy concerns, budgetary silos, and the simple inertia of institutions that weren't designed to share. AI, paradoxically, can help break this deadlock — but only if cities commit to curated, privacy-preserving data pipelines that make regional intelligence sharing safe enough to actually happen.
The traffic situation that a City is trying to solve cannot be fixed if the cities around it aren't solving theirs too."
04 A Path Forward: Start Small, Scale Smart
The speakers closed with practical wisdom that is easy to overlook in the excitement of grand technological visions: start with bounded, high-impact projects where success is measurable and the data requirements are manageable.
This incremental approach is not a compromise, it is strategy. Each small win builds institutional trust, generates the training data that makes AI smarter, and creates the cross-department (and cross-city) relationships that larger collaboration ultimately requires. The cities that will lead in Physical and Agentic AI won't necessarily be the ones with the biggest budgets; they'll be the ones with the most disciplined starting points.
The fireside chat was a timely reminder that the future of cities isn't being written in a single sweeping policy or a billion-dollar contract. It's being sketched — carefully, collaboratively — at the intersection of data, infrastructure, and the will to share both. AI gives us tools of extraordinary power. The harder work, as always, is deciding to use them together.
Comments
Post a Comment