Beyond Limits: Cities Large and Small Put AI to Use

As artificial intelligence has continued to rapidly evolve in recent years, many government agencies have started to address and deploy the tech, both at the policy level and by testing use cases with potential to improve their work and services. Many of the high-profile uses of government AI, however, have so far been concentrated at the state and federal levels.

But experts say that local gov tech leaders are starting to catch up. While cities and other local jurisdictions may not have the financial resources or the diverse pools of tech talent that larger levels of government do, cities are both more agile as well as closer to their constituents and their constituents’ problems. This creates less risk along with a wider range of services on which to test AI.

A number of interesting AI use cases have started to emerge at the local level of late, said Lynn Overmann, executive director of the Beeck Center for Social Impact and Innovation, which is housed at Georgetown University in Washington, D.C. The Beeck Center is a nonpartisan, vendor-neutral group, and part of its mission is to help government organizations assess the pros and cons of emerging technologies — technologies such as AI.

“As cities are getting savvier and savvier with AI,” Overmann said, “I think people are at the stage where they’re zeroing in on, ‘How do I make this work? And how do I make it work while protecting my populous and minimizing risk?’”

This has led to use cases in places like San Francisco, where the city has teamed up with Stanford University to create an AI tool that can effectively analyze the municipal code, which is more than 16 million words and rather cumbersome for any human to read, let alone parse in a meaningful way. So, technologists in San Francisco used the tool to find sections of the code that just aren’t relevant anymore, like one that required placement of newspaper racks in the city’s streets, Overmann said.

But San Francisco is, perhaps, exactly the sort of steadily innovative city that one would expect to already be using AI. What’s starting to change now is that the technology is also being put to use in other jurisdictions across the country.

LEBANON, N.H.

Lebanon, N.H., is exactly the sort of smaller, more agile jurisdiction that experts say can move faster and find new public-sector uses for AI.

With a population just shy of 16,000, the city is nestled in the western part of its state, adjacent to the border with Vermont. Melanie McDonough is Lebanon’s chief innovation and AI officer, and she agrees that being a smaller organization has helped the city move faster with the new technology.

“At smaller organizations we tend to wear a lot of different hats,” McDonough said. “The same person might be handling your IT, your data privacy and your data governance.”

And as a result, that person can make decisions more quickly and also scale projects more effectively. While Lebanon is landlocked, McDonough drew a maritime metaphor to explain the situation: Deploying new tech for existing government work can be like changing the direction of a ship, and a big ship takes much longer to redirect than a smaller, swifter boat.

For Lebanon, this has so far led to a wide range of use cases, from helping internal staff write better emails to conducting deep research on the policies of other jurisdictions with the potential to improve the city.

One clear use of AI for the city has been helping firefighters and EMTs write reports. McDonough said the city identified this as a pain point, noting that people who choose those professions do so to help and save lives — not to carefully document incidents in ways that capture every last mandated detail. But that’s where AI can help.

While the city can’t use the technology to write reports in their entirety, it can use AI to analyze the reports that humans write and instantly let them know when they’ve missed legally mandated criteria, which saves time over a human having to do it. This use case, she noted, has also improved firefighter and EMT morale.

That’s just one example. The city has also used something called vibe coding — where AI helps with the specifics of coding based off a prompt — to make its strategic plan accessible and searchable online. McDonough, who is not a coder, was able to undertake that project herself, although she noted that it is only because she knew what was needed inside and out.

“What we’re doing is getting tools to the people who actually know what the problems are, but didn’t have the skill set prior to that to deliver coding of that caliber,” McDonough said. “It would have been something we would have had to hire out.”

And while there are and should continue to be conversations in Lebanon about whether uses of AI are responsible, McDonough said the city tries to stay focused on how it can be used to better serve residents, which is, after all, government’s chief function.

“We’re so close to the feedback,” she said. “People are walking into our City Hall and telling us this is what we like, this is what we don’t like … we tend to hear directly whether it’s working or not working.”

And the learnings in Lebanon can also help other levels of government, too. McDonough is a member of the New Hampshire AI Task Force, a state-level group that works to make New Hampshire AI ready.

AUSTIN, TEXAS

While Lebanon, N.H., is a great example of how AI is being used in smaller cities, there are larger cities that are doing interesting things with AI of late, too.

With a population of nearly 1 million residents, Austin, Texas, is very different from Lebanon, and it’s doing different things with AI-powered technology. Divya Rathanlal is CIO of Austin’s water utility, but until late September she was the city’s interim deputy CIO. She said that like many other cities, Austin is currently on the cusp of moving from AI pilots to fully deploying the technology as an efficient part of its services and workflows.

But that leap first requires a few things: identifying a need, undergoing testing and fostering trust from residents or from the relevant city staffers who will use the new systems.

“We don’t want to develop technology for the sake of technology,” Rathanlal said. “We want to make sure it provides value to our residents.”

With this in mind, there is a rich set of AI pilots underway in Texas’ capital. One example includes a smarter permitting system, which is not entirely unlike the first responder report use in Lebanon.

In the past, when an Austin resident applied for a construction permit, that application had the potential to get stuck in a lengthy queue if there were any details missing. Austin’s new AI tool can now quickly scan residential construction plans and let people know in an instant if they’re missing any attachments or other necessary information, subsequently kicking it back to the applicant and reducing the time before they ultimately get approval.

If that project goes well, Rathanlal said it could be extended to commercial or multiresidential construction permitting, too.

Another project in Austin involves an AI assistant to help people file nonemergency police reports. Instead of a person calling 911 and creating a backlog there, a caller without a pressing emergency would work with an AI assistant that can collect basic information. If necessary, that assistant can also let a human officer know to contact the caller for additional info.

Finally, the city is working with Austin Energy — a utility — and the private company AI Fire on a program that uses AI-powered cameras to continuously monitor the air in and around the city for signs of heavy smoke. If it sees an area of concern, it then contacts emergency responders to check it out.

Part of what’s made this work possible is a partnership between the city and its local universities, particularly the University of Texas-Austin. Each year, the city and the school come together to review potential use cases, exploring whether they can partner up on using tech to address them.

“It’s always business-driven and technology-enabled,” Rathanlal said.

Another collaboration has helped make Austin’s work possible as well. Austin was an early member of the Government AI Coalition, which was founded in San Jose, Calif., and essentially created a network of cities interested in learning about AI uses from each other. Rathanlal has described that membership — as well as other learning partnerships with cities — as absolutely vital to advancing her jurisdiction’s use of AI.

KANSAS CITY, MO.

But Austin, of course, is not the only city embracing collaboration with other jurisdictions. In Kansas City, Mo., Chief Digital Officer Andrew Ngui also speaks highly of the value of working with his peers on how best to harness the rapidly evolving technology in responsible ways that provide value to residents.

Like Austin, Kansas City was one of the early members of the Government AI Coalition. In July, Ngui even wrote an op-ed for the organization about the potential of AI to help government agencies deliver services that are human-centered.

He served as a member of a working group around how to use AI for translation services. Ngui said that the work was extensive and productive, involving more than 700 government agencies and 2,000 agency staffers. It yielded a data set of 27,000 government words and terms that all cities or other local agencies can now download, translated across 14 different languages.

A major AI success story for Kansas City is the creation of the multilingual chatbot Maya within the city’s BizCare office, which helps residents start their own businesses. The chatbot is equipped with more than 70 languages and offers 24-hour availability. If a resident has a specific question about starting a business at 3 a.m., they can get that answered.

Like many other cities, Kansas City is also still working to hone its policies around AI use, enhance existing AI projects like chatbots and make sure that its staffers are seeking out ways to use AI that will make their workflows smoother. Ngui stressed that AI implementation is not just an opportunity to serve Kansas City’s businesses and residents, but it is also an opportunity to help the city’s staff, emphasizing the tech’s role as a professional development tool.

Looking ahead, Ngui — as well as the civic technologists in Lebanon and Austin — is bullish on AI’s potential, while at the same time stressing the importance of continuing to discuss its responsible use. That conversation is an important part of creating a culture where residents and staffers alike both understand and want to use AI.

The future, of course, is difficult to predict, and Ngui noted that the future of AI use in local government will not be predicated on the work that any of these cities are currently doing, “but more importantly on the imagining and dreaming of the use cases we haven’t even come to think about.”

This story originally appeared in the Fall 2025 issue of Government Technology magazine. Click here to view the full digital edition online.