Minnesota’s legislative session moves fast— thousands of bills are introduced in a matter of weeks. Each year, agencies rush to sort through the long list of legislation to find and analyze bills that could have a big impact on their operations, policies, and budgets. During this critical period, time is of the essence. For agencies like Minnesota’s Department of Revenue (DOR), the manual process of finding and analyzing legislative bills takes a lot of time and effort. Each session, DOR staff spends thousands of hours searching for bills that are relevant to their agency.
| Year added | 2025 |
|---|---|
| Status as of Spring 2026 | Deployed |
| Primary Agency | Legislative Branch |
| Approach and Solution | Minnesota IT Services and the Department of Revenue built an AI-powered bill-review system on tools the state already owned — SharePoint, Microsoft Azure, and a secure government instance of OpenAI — with no new infrastructure or vendor contracts. A web scraper monitored the Minnesota Legislative Revisor’s website and captured newly introduced bills in real time. These bills were queued in SharePoint, then routed through Azure Functions where a custom OpenAI prompt analyzed, tagged, and summarized each bill. Results, including relevance scores and summaries, were returned to SharePoint for immediate staff review. The system processes bills in parallel at speeds up to 100 bills per minute. ChatGPT evaluates each bill’s content for relevance, statutory changes, and impact across divisions. |
| Who it Serves | Government staff |
| Area of Impact | Legal & Regulatory |
| AI Technology Type | Generative AI |
| Level of Human Oversight | AI-assisted decision (human review required) |
| Impact / Outcomes | The system processed more than 6,500 bills at 99% classification accuracy, at speeds up to 100 bills per minute. Staff reported that reviews that once took two to three hours were completed in about ten minutes, saving hundreds of hours per session and freeing policy experts to focus on substantive analysis rather than sorting. Built in 90 days with no new funding, the pattern is now drawing interest from other Minnesota agencies tracking health, education, and transportation legislation. |
California developed a statewide platform for Californians to share details on how AI is affecting their work and our economy, and what the state should do about it. The feedback will be used to inform and guide state policy leaders on this new technology.
The Morongo Band of Mission Indians built a proprietary chatbot trained on decades of Morongo legislative history that serves as an AI-driven legal repository. Officials now get answers in 10 to 15 seconds instead of 10 to 15 minutes.
The Department of Justice deployed Lexis+ AI, a commercial generative AI legal research tool, to streamline research workflows for its attorneys. The tool was identified as an operational AI system in the New Hampshire Department of Information Technology's July 1, 2025 annual AI systems report.
The court launched an AI communications platform as a pilot with Amazon Web Services using CloudHesive ConnectPath. The system includes a multilingual virtual assistant that gives residents 24/7 access to court information via text messaging in English and Spanish, delivering real-time links to virtual hearings, court schedules and notices, FAQs, instructional videos, documentation, and contact information for public defenders and prosecutors. The county also plans to add self-service kiosks in participating town halls so residents can pay fines locally or online rather than traveling to the Mays Landing courthouse.