My big project this summer has been obtaining guest parking permits for my new house. It took two trips to the ward office, an hour in an automated phone tree and a detour to a state facility. The passes finally hit my mailbox last week, several aggravating months after I started.
I’m not alone in my frustrations accessing city services. In the Chicago Office of the Inspector General’s audit of Chicago’s 311 system, residents described their experiences interacting with the city as a “black hole,” with no way to track and follow service requests.
Businesses, meanwhile, have sprinted to adopt artificial intelligence to better serve customers. Voice agents answer calls instantly. Document models process in minutes what once took weeks. Back-office AI verifies identity and schedules appointments without human involvement. Companies have transformed their products and services over the last 36 months, but American cities have not yet followed.
More Top Picks Best Sake Warmers
Chicago’s next mayor can lead the way: Draft the AI playbook every city in the country will copy and transform how local government delivers for residents.
How do I know this is possible? It’s my day job. I built a venture-backed company in Chicago and sold it to a publicly traded tech company. Today, I run an AI startup that puts this technology to work inside large enterprises. Every week, I watch AI do for a company what Chicago can’t do for its residents.
Other governments have already gotten to work. The British government is hiring AI specialists to work directly on public problems. New York just launched Mayor Zohran Mamdani’s “PIT Crews,” 30 in-house technologists building digital tools for city agencies.
Start with 311. A modern voice agent can resolve routine requests instantly, at any hour, in any language. Complex cases can be referred to a person, with conversation history attached. Midland, Texas, a city a fraction of Chicago’s size, eliminated 7,000 missed calls by deploying a municipal voice agent. We should take the same approach.
The opportunity goes well beyond phone calls, however. Large language models now match or exceed human accuracy on routine document review and could easily take on the work of issuing city permits and licenses.
Take my guest parking permits. Today, the clerk’s “online” process means completing a form, scanning and emailing verification documents, waiting four weeks for human review and then waiting two more for the passes to arrive by mail. Let AI verify proof of residency, flag only the unusual cases for staff review, and have the passes printed and mailed the same day.
Or consider the Department of Buildings, where a few dozen inspectors chase tens of thousands of complaints a year — a system investigative reporters have linked to preventable fire deaths in buildings whose violations were recorded but never resolved. Stockton, California, now uses AI to scan for code violations proactively and rank them for inspectors; its pilot surfaced 4,000 violations in five days. Let AI read every open case in Chicago overnight and put the life safety risks at the top of an inspector’s list by morning.
Deploying technology has pitfalls, and Chicago has a fresh reminder of the dangers. Cook County’s decadelong effort to modernize its property tax system ballooned in cost from $30 million to $63 million, led to delayed tax bills and at its peak left roughly 83,000 property owners waiting on $186 million in refunds.
Two lessons stand out from the Cook County inspector general’s assessment of the fiasco. First, hiring and procurement rules make it nearly impossible to bring technical talent in-house. As a result, agencies lack understanding of how technology works. Second, governments prefer a single vendor and require customization, which takes several years to implement. These all-or-nothing contracts disincentivize iterative development and allow vendors to hold their customers hostage.
Here’s how the next mayor can avoid those mistakes.
First, hire a small number of elite engineering and product leaders who believe in the civic mission. Hiring an army of in-house engineers to build and maintain applications is impossible in a world where AI researchers command eight-figure pay packages. Instead, we should focus on hiring a handful of people capable of evaluating vendors and managing them to outcomes, and building high-impact applications with modern development tools such as Claude.
Second, work with startups, not legacy vendors. Use that talent to run dozens of pilots with startups in parallel, starting with high-volume, low-risk services. Pay for outcomes, not licenses. Contracts should reward resolved calls and cleared backlogs, and delivery risk should sit with the vendors. My company doesn’t sell to government, but many of my peers would jump at putting the city of Chicago on their logo wall. The best startups will compete hard for this work.
The next mayor can make Chicago the national model for what AI can deliver for residents. Nobody should wait months for parking passes.
Tom Dowling is co-founder and CEO of FullSeam, a Chicago-based AI startup, and the founder of TaxProper, which was acquired by Opendoor in 2022. He is a commissioner of the Illinois Student Assistance Commission and previously advised Mayor Lori Lightfoot.
More Top Picks Best Dip Belts For Weighted Dips
Submit a letter, of no more than 400 words, to the editor here or email [email protected].