Local governments could deploy AI for good. Here’s how

When considering AI’s impact in cities, many residents and government officials envision a dark future of unbridled surveillance, hollowed-out city halls and unaccountable bots calling the shots based on biased training data. We, on the other hand, embrace a much more optimistic vision. With ambitious local leadership, AI, and especially the coming wave of agentic AI, can offer a profound opportunity not only to make government services more efficient but also to transform how cities fulfill their end of the social contract. 

As long-time public servants and champions of government innovation at our respective universities, we understand the challenges local governments face, including tight budgets, aging infrastructure and dissatisfied residents accustomed to the speed of Amazon and personalization of Spotify. Most cities still run on a century-old operating system built on bureaucracy, paper files, agency silos and rigid hierarchy. Agentic AI offers a unique opportunity to redesign how cities work, a model we call the “Agentic City.”

Agents, city employees, and citizens working together

Imagine a city administration where the complexity of navigating government bureaucracy is offloaded to intelligent agents—so routine tasks happen flawlessly, and even complex ones feel simple. A mother reports a broken sidewalk near her child’s school, snaps a photo, and sends it to the city. An AI agent classifies the problem, routes it to the right crew, tracks progress across agencies, proactively updates her until the work is done, and alerts others to similar risks nearby. Imagine a city that fixes pavement cracks before they become potholes, changes street lamps before they burn out, and repairs water lines before they leak.

Yet even these dramatic improvements will only constitute steps in a transformation. These tools help reform-minded mayors adopt a system approach that sidesteps the strong headwinds often confronting business reengineering, including efforts to integrate agency functions or disparate data systems. Transportation officials no longer need to tweak signals; an AI traffic agent can balance safety, travel time and emissions. An Agentic City will be one in which agents, public employees and residents work together.

Ultimately, all city services will be personalized as residents use an “agentic front door” to state their goals (“want to open a barber shop at 10^th and Main”). Agents will walk users through the process or even complete those tasks for them. At the same time, a human monitors the results, troubleshoots and takes on difficult or unusual cases. In fact, this city offers preemptive housing vouchers, rental assistance, and property tax relief to those who qualify, obviating the application maze entirely.

A systemic approach

Getting there will require strong leadership to overcome gaps in imagination, skill deficits, and employee anxiety, compounded by the complexity of ensuring that AI changes comply with democratic values. Local leaders will need to take a systematic approach, crafting a powerful narrative of the service benefits while using their political and legal skills to negotiate with the city council, union, and employee leaders.

AI-driven transformation requires a leadership team supported by academic and other local experts who understand the city’s technical capacity, legal and data limitations, and that stretches the imagination of a bureaucracy accustomed to existing processes. That team should establish a pathway for opportunities for both employees and residents, including the agentic front door, repetitive functions that can be outsourced to AI, and more time for staff to take on higher-value purposes: investigating root causes, engaging communities, and exercising judgment. Third, the leadership team should promote the incorporation of agentic capabilities that help employees identify patterns and causes of recurring problems by making data more easily accessible.

Municipal workforces, both union and nonunion, represent a key stakeholder. Mayors need to be clear that AI will complement, not replace, the workforce. An Agentic City initiative would include outreach to labor to set the parameters of a new bargain in which workers, armed with data insights, increase productivity and share in the benefits through pay increases. Data literacy training and a data governance framework should also be essential components. Freed of repetitive tasks, public employees can focus on higher-value work.

The data foundation

Addressing these concerns responsibly begins with the system’s foundation: the data. Cities must invest in data pipelines that are not merely machine-readable but machine-understandable—structured with rich metadata, shared ontologies, and business-logic context—so that both humans and AI agents can interpret meaning, constraints, and appropriate use. Emerging approaches such as Model Context Protocols (MCPs), which standardize how AI systems access structured data and operational tools, represent a promising step in this direction by helping agents understand not only what data exists but also how it should be used. An agent that can “see” a permit record but not understand the regulatory framework, eligibility rules, or data quality limitations behind it will act inconsistently and require constant human correction. Machine-understandable data reduces that friction and makes agentic systems more reliable, transparent, and scalable. In short, the foundation of an Agentic City is not just smarter algorithms, but smarter data architecture.

Implementing an agentic city hall presents substantial challenges. However, now is the time to lead, as mayors cannot afford to maintain the status quo or wait for the AI tsunami. Going forward presents challenges as well. Doing nothing poses a greater risk than getting started, and the evidence will be a city that, through more meaningful work for its employees, becomes more responsive to its residents.

View the full article