How San José trained 1,000 city employees to build their own AI tools

San José’s in-house artificial intelligence training effort has reached a major milestone, with more than 1,000 city employees completing the program since its launch in 2024 and creating AI tools that the city says are already reducing administrative burdens across departments.

The city announced this week that its AI Upskilling Program, launched in partnership with San José State University, has trained roughly 15% of the municipal workforce. The voluntary program includes self-paced courses and a 10-week cohort-based training track where employees design AI tools tailored to their roles.

Stephen Caines, San José’s chief innovation officer, said that the government adopted a “bottom-up” approach that encourages employees to identify challenges and develop solutions themselves. The city aims to expand participation to 2,500 employees, or about 30% of its staff, by June 2027.

“Bottom up is really about asking the people on the front line of the work, where do they see value, where do they see service gaps, and where are those opportunities for improvement,” Caines told StateScoop in a recent interview.

As a result of that approach, Caines said demand has consistently exceeded available seats.

“It’s typically oversubscribed,” he explained. “We’ve even seen people come back to take the course again multiple times.”

San José is one of dozens of cities across the U.S. that are rapidly rolling out generative AI training for public servants to boost efficiency and safety.

In February, Washington, D.C., became the first major U.S. city to mandate AI training for all employees and contractors. The free training program covers prompting basics, responsible innovation and recognizing deepfakes. San Francisco city employees also have access to AI literacy courses through a government portal.

Caines said the city’s AI governance framework is embedded in the program. Employees build tools using approved enterprise AI platforms and city-managed accounts, eliminating the need for separate reviews in most cases.

In the two years since it launched, the initiative has produced AI applications across multiple city departments, including a tool that automatically verifies emergency vehicles are equipped before deployment, a system that reviews contractor submissions to identify missing information and drafts response emails and an AI assistant that evaluates projects that further San José’s goal of achieving carbon neutrality by 2030, among others. The city is also exploring future uses for AI, including connecting large language models to municipal open-data systems through model context protocol servers.