Key takeaways
- NASCIO's 2026 State CIO Survey found 82% of states have AI projects in production, and 35% (about 18 states) run enterprisewide AI in the executive branch.
- AI requests now come from operational agencies. In California, a sanitation district, a police department, a building department and a housing authority all issued AI-related solicitations within 30 days.
- Build AI prospect lists by department and use case, not only by state CIO, and keep central IT and policy offices in view because they set the rules.
- Expect to explain how your product uses AI. Requiring vendors to disclose AI use is an emerging and growing practice among states.
For the last few years, selling AI to government mostly meant one conversation: the state CIO's office, an innovation team, maybe a pilot. That conversation still matters, but it's no longer the only one.
New survey data from state CIOs and a burst of solicitations in California show AI moving out of pilots and into the departments that run sewers, permits, policing and housing. If your AI prospect list starts and ends with central IT, you're missing a growing share of the people writing requirements.
What the state CIO survey shows
The National Association of State Chief Information Officers (NASCIO) released its 2026 State CIO Survey on Sept. 28 at its annual conference in San Diego. As reported by Government Technology, the AI findings point to a technology moving past experiments:
- Nearly every responding state has AI pilots or proofs of concept, and 82% have projects in production.
- 35% of states, or about 18, have reached enterprisewide AI projects in the executive branch. This is the first year the survey measured it.
- Almost 30% say they are using agentic AI, meaning systems that can take actions on their own, and another 35% say they plan to.
- 98% have enterprise policies and procedures on AI development and use.
The move to production matters for sellers. Pilots are usually owned by a central team. Production systems belong to the program that uses them, and that program's leaders have their own budgets, their own problems and their own say in what gets bought.
California shows where the requests come from
Governor Gavin Newsom signed Executive Order N-5-26 on March 30, giving the California Department of Technology and the Department of General Services 120 days to set new AI vendor requirements and expand employee access to vetted generative AI tools. According to the law firm Pillsbury, the order uses the state's buying power to require AI vendors seeking state contracts to show safeguards against illegal content, harmful bias and civil rights violations.
The 120 days ended on July 28, and solicitations followed. NationGraph counted nine AI-related RFPs in California in the 30 days ending Aug. 27, about 3.2 times the state's trailing monthly average. The issuers were not central IT:
- Sanitation Districts of Los Angeles County: an AI/machine learning solicitation for sewer-pipe analysis.
- San Diego Police Department: an AI report-writing tool.
- Los Angeles Department of Building and Safety: an AI-powered permit-check assistant.
- Alameda City Housing Authority: consulting on AI governance policy.
Smaller bodies are moving too. The Nevada Irrigation District's August meeting minutes show it drafting a formal AI governance policy.
Build the prospect list by department and use case
The practical change is in how you build your list. Instead of one row per state, think in terms of use cases and the departments that own them:
- Start with your use cases. Write down the two or three problems your product solves best, such as report writing, document review, inspection analysis or resident service.
- Map each use case to the departments that own it. Report writing points to police departments and sheriff's offices. Pipe inspection points to sanitation, sewer and water districts. Permit review points to building and planning departments. Our list of the largest police departments is one starting point for public safety.
- Look for evidence of intent. RFPs are the obvious signal, but board minutes, budget documents and draft AI policies show up earlier. Our guide to government buying signals covers where to find them.
- Keep central IT and policy offices in view. The CIO's office and state procurement still set standards, approved tools and contract vehicles. A department may want your product and still need it to clear those rules. Our state AI policy tracker shows where each state stands.
Don't skip special districts. Sanitation, water and irrigation districts buy on their own and rarely show up on lists built around state agencies.
Be ready to explain your own AI use
NASCIO's survey found that most states have use case inventories and special procurement requirements for AI, and it describes requiring vendors to disclose their use of AI as an emerging and growing practice. California's order goes further by asking vendors to show safeguards.
Before you reach out, have plain answers ready: what AI your product uses, what data it touches, where that data goes, how you test for bias and errors, and how a person stays in the loop. A one-page summary helps department heads who must explain the purchase to their own IT and legal reviewers. We covered similar ground for federal bids in our post on AI governance in federal proposals.
Keep the outreach low-pressure
Many of these buyers are new to AI purchasing. A police captain or district engineer may have a clear problem but little experience writing AI requirements. Lead with their problem, not your model. Ask how they handle the work today, whether AI is part of their 2027 planning and who else needs to weigh in. If the answer is "not yet," ask when it makes sense to check back, and then do exactly that.
Leadership changes are another reason to build department-level relationships. With 18 states guaranteed a new governor after this year's elections, many states will have new CIOs in 2027. Operational leaders who own a problem often stay in place through those transitions.
How KSG helps
Kensington Sales Group's market intelligence team builds prospect lists by use case and department, from police departments to sanitation districts, and tracks the solicitations and signals that show who is ready to buy. Our SDRs then set meetings with the operational leaders who own the problem, with room for them to say "not now."
Sources: Government Technology, NASCIO 2026 State CIO Survey: Finding a Path to AI Maturity (September 28, 2026); NationGraph, California's AI Procurement Clock Ran Out and the RFPs Started Pouring In (August 27, 2026); Pillsbury, California Executive Order N-5-26: Procurement Power as AI Policy (May 14, 2026).