State governments are moving agentic AI out of pilot programs and into the services residents use every day. The shift follows the pandemic, when the Kansas Department of Labor's 1970s-era system nearly collapsed under a hundredfold spike in daily calls. The rebuild that followed shows what is at stake when legacy technology fails - and what AI can deliver when it replaces it.
The Kansas labor system was available only 65% of the time, calls took 30 to 40 minutes to resolve, and training a new employee took eight weeks. Now, a cloud-native, AI-enabled platform built in 29 months runs 24 hours a day with no downtime. About 90% of people seeking help can be served online, and training time has dropped to two weeks. Call transcription and auto-adjudication are handled by machines, freeing workers to focus on cases that need human judgment. This push toward AI Agents & Automation in government services is shifting how agencies operate, not just how they process data.
A 50-year-old system gets a 29-month rebuild
Kansas Secretary of Labor Amber Shultz told Route Fifty the state built for decades, not just for today. "We're really failing individual citizens, businesses, our stakeholder groups, organized labor. It's not just one person; we're failing everybody," she said. Alec Chalmers, AWS' director of educational and government technology, said AI will allow governments to modernize legacy applications that are in some cases 30 and 40 years old.
Governance first, agents second in Tennessee
While Kansas rebuilt from crisis, Tennessee is preparing methodically. State CTO Jerry Jones built governance and workforce structures before deploying any AI tools, according to StateTech Magazine. The state created an AI advisory council in statute and a review committee that requires agencies to justify business value, funding and data readiness before launching pilots. "Everybody has a great idea for AI. You have to have a way to filter that," Jones said.
With guardrails in place, pilots are underway in legal, citizen services, IT operations and back-office workflows. One uses AI to retrieve and redact information for public records requests, work that previously required significant manual effort. Another is a statewide chatbot that helps residents navigate benefits eligibility. Jones stressed starting where the risk is low: "Start in the back office. If it goes sideways, nobody's going to die." He said the state is moving from systems of record to systems that act, and that agentic AI is coming, but the foundation must be built first. This AI for Government focus is spreading as more states adopt governance-first frameworks.
AI in the courtroom, pharmacy, and permit office
Other governments are putting AI to work in specific services. In Allegan County, Michigan, public defender Chad Catalino deployed JusticeText to search and transcribe body cam footage. Case clearance rose 8 to 9% after a year, and seven statewide systems now use the tool, including Montana, Iowa and Kentucky.
Utah became the first state to allow AI-driven prescription renewals, with physician sign-off required on the first 250 interactions. In Lancaster, California, Mayor Rex Parris deployed an AI permitting tool that reviews draft applications against 18,000 pages of updated building code. "Governments shouldn't be adding to the cost of things. They should be making them cheaper, more affordable, and AI really helps us do that," Parris said. North Las Vegas became the first jurisdiction in Nevada to offer real-time AI translation at public meetings in virtually any language. Forty percent of the city's population speaks a language other than English, and the tool gives many residents a voice they did not have before.
Why this matters for government agencies
The stories from Kansas, Tennessee, and other states carry a clear signal: the move to agentic AI is not waiting for perfect conditions. Kansas rebuilt a broken system in 29 months after a 50-year failure. Tennessee built governance structures first to make sure AI projects are funded, justified, and ready. The deployments in Michigan, Utah, California, and Nevada show that even targeted AI tools can reduce case backlogs, speed up permits, and expand access to public meetings. For agency leaders, the lesson is that waiting to modernize legacy systems carries its own cost - one paid in delays, citizen frustration, and staff burnout. The technology is moving from the back office to the front counter, and the states that are seeing results are the ones that paired bold rebuilds with practical guardrails.
Your membership also unlocks:
