Since September 2025, public professionals have come together to learn from one another how to use AI to improve government.
Through more than 20 InnovateUS workshops, experts examined real cases—from hiring reform and benefits enrollment to democratic engagement and data-driven management—that show how public servants are redesigning systems with people, not just products, at the center.
Success with AI follows a simple sequence:
When governments start with people, such as training staff, listening to users, and clarifying outcomes, AI enhances the end result. When they skip straight to tools, the results fall flat.
As Harvard professor Danielle Allen explained: “You can generate decisions faster without participation, but they’ll be thinner and more brittle.”
Five Ways Public Sector Teams Are Redesigning Systems Using People → Process → Technology
Modernizing St. Louis Hiring Systems
Watch: Hiring Reform with Humans at the Center: Lessons from St. Louis and San Francisco
In St. Louis, public hiring was paralyzed by outdated workflows. Some roles took up to 12 months to fill, and many departments still used typewriters. Adoption stalled despite having access to AI tools because the systems weren’t ready.
Krizia Lopez led the transformation by retraining 700 staff, rebuilding hiring processes, and fostering buy-in across departments. After that groundwork, the team reintroduced AI-led resume screeners and scheduling bots.
“You don’t need to chase the next shiny thing. Work with what you have, for who you serve.”
— Krizia Lopez, Vice President, Government Innovation, Work for America
As a result, time to hire dropped to just 2–3 months, and the number of positions filled annually rose from around 500 to over 880. Public services became equitable and more efficient because people trusted the process again, and tools could finally do their job.
Colorado Operations Built Around Outcomes
Watch: A Model for State Government “Operators” - Colorado’s Governor’s Office of Operations
Colorado’s state government had a performance problem: only 52% of priority goals were being met, and departments lacked alignment.
Jesse Marks and Emily Miller changed that trajectory by convening cabinet leaders to define shared outcomes such as healthcare savings, universal preschool, and crime reduction. With those goals, they rebuilt workflows, launched real-time dashboards, and introduced virtual agents to handle up to 40% of unemployment calls.
“You can’t build trust in a moment of crisis. You need relationships ahead of time.”
—Emily Miller, Deputy Director of Operations, Colorado Governor’s Office of Operations
With the systems realigned, the state’s performance transformed. Unemployment call wait times dropped from 36 to 11 minutes, agency alignment improved, and primary goals were achieved, including $2 billion in healthcare savings, 70% preschool enrollment, and a 26% drop in auto theft.
Embedding Human Navigators to Streamline Benefits Enrollment
Watch: Leveraging Artificial Intelligence to Enroll in Federal and State Benefits
Patients can be eligible for public benefits in clinics and emergency rooms, but few receive them. Complex and bureaucratic processes stand in the way.
Dr. Alister Martin tackled this by embedding “human navigators” directly in clinical settings. Once those frontline relationships worked, his team layered in AI to prefill forms, check eligibility across 20+ programs, and cut through bureaucracy.
“AI doesn’t deliver trust. People do. But AI can remove friction.”
—Alister Martin, ER Physician and Founder, A Healthier Democracy.
This pairing of trust and technology paid off: one mother received $1,600 during intake and over $20,000 in her first year of enrollment. Navigators became five times more efficient without losing the human connection that made their work effective.
Participatory Planning in Hamburg
Watch: Defining Problems with the Public
Hamburg’s DIPAS platform began with a decade of participatory governance led by Claudius Lieven’s team, 130 open-source civic processes built on anonymity, transparency, and feedback loops.
Because the public already trusted the system, Hamburg could confidently use AI to analyze 11,000 comments on an urban plan. As a result, they uncovered that residents were more concerned with cultural identity than infrastructure, something the original question hadn’t asked.
“When you ask the public the wrong question, they still give the right answer. You just have to be listening.”
—Claudius Lieven, Project Lead, DIPAS, Hamburg
That insight surfaced in days instead of weeks, allowing planners to rewrite engagement prompts in line with public priorities and encouraging residents to return and participate in future processes. AI helped scale a system already built for trust and responsiveness.
Language Access Grounded in Dignity
Watch: Beyond Translation Workshop
In New Jersey, Barbara Niveyro and Ruthie Nachmany spent eight months co-creating new unemployment materials with native Spanish speakers, rewriting jargon-heavy forms in plain, culturally relevant language. Then they added AI translation to support scale and maintenance, not as a substitute for human-centered design.
“We designed it like your grandmother would explain it to you.”
—Barbara Niveyro, Bilingual Content Design Lead, NJ Office of Innovation
The changes dramatically improved access: Spanish-language users could complete the process in just 25 minutes instead of four hours, reaching parity with English users. And because the system reduced errors and paperwork, the state saved $250,000 in processing costs.
Five Lessons That Keep Coming Up
- Start with the affected people. Before using AI, the workflow, trust, and training must be redesigned. It’s not just tools, it’s systems.
- Define what success entails. First, set your outcomes. Build dashboards later. AI can’t replace an agreed-upon strategy.
- Fix points of friction by using AI to speed things up, but without excluding people. Keeping humans in the loop is fundamental every step of the way.
- Civic input without follow-through feels extractive. Transparency builds legitimacy.
- Communication is infrastructure. If people don’t know what changed, or why, it doesn’t matter how good the tech solves the problem.
As Jill Abramson reminded us: “Public trust is earned by making change legible.”
You don’t need a ten-year roadmap or 700 staff to start.
But you do need the sequence: People → Process → Technology. When we create the foundation first, AI works for democracy, not the other way around.
What’s Next?
The conversation isn’t over. We’re just getting started. Here’s what you can do:
- Explore the recordings. All sessions are available online at innovate-us.org. Watch one over lunch, share it with your team, and use it to spark new thinking.
- Apply the frameworks. Each workshop comes with resources, from participation platform guides to language access toolkits.
- Join the next conversation, more sessions are available every week. Check back at InnovateUS, follow @InnovateUS and the @Burnes Center, or subscribe to updates to stay in the loop.
If you work in government or alongside it, there’s never been a better time to learn from each other. Democracy isn’t automatic. Let’s design it together.