San José says it has quietly pulled off a big tech pivot inside City Hall: about 1,000 city employees have now finished an in-house AI upskilling program and are building custom tools to speed up everything from 311 triage to permit reviews and emergency readiness. City officials and program graduates say the training is stripping routine drudge work out of the day and giving staff more time for higher-value projects. The broader experiment is testing whether municipal employees, rather than outside vendors, can safely weave generative AI into everyday city operations.
In a press release from Mayor Matt Mahan's office, the city said its AI Upskilling Program, launched in September 2024, has already graduated multiple cohorts and reached roughly 15 percent of the municipal workforce. That release also estimated that early cohorts have saved between 10,000 and 20,000 staff hours and avoided about 50,000 dollars in consulting costs. Mayor Matt Mahan promoted the milestone on Facebook and linked to coverage of the program in a Facebook post.
The city’s IT Training Academy runs a 10-week, cohort-based AI track in partnership with San José State University. The curriculum blends short self-paced modules with hands-on projects tailored to each department’s needs. According to the Academy, participants often claw back more than an hour a day on chores like drafting memos, reviewing grants, and summarizing long reports, and some staff have built custom assistants using ChatGPT to automate routine steps. Graduates present their final projects to City leadership so promising tools can be adopted and governed inside official systems.
How the training works
The program leans on a bottom-up model that asks frontline employees to bring real work problems to class and design AI helpers to solve them, using enterprise accounts and approval checkpoints that keep models and data under city control, as reported by StateScoop. Trainers work with San José State and city IT staff to map projects to San José’s generative AI policies and privacy safeguards, instead of letting individual employees experiment freely on public systems. City leaders say this approach is meant to make useful projects easier to scale without creating fresh privacy or procurement headaches.
Early wins: 311, contractors and fire readiness
The mayor’s office and city communications teams point to a handful of early success stories. One transportation staffer used a custom assistant to secure and pivot on multimillion-dollar EV charging grants, while an IT analyst built a tool that parses the 311 “Other Issues” field into recurring themes, saving hundreds of staff hours a year. City materials also describe assistants that automatically check whether emergency vehicles are properly stocked before deployment, along with systems that flag missing information in contractor submissions, steps the city says cut down on back-and-forth that can slow infrastructure work. According to a city release, those examples are meant to show how relatively modest, staff-built automation can shorten the time it takes to move projects forward.
Governance and what’s next
San José has woven governance into the training from the start: employees are required to build tools on approved platforms and to follow published generative AI guidelines that outline rules for data handling, model limits, and public records obligations, according to the city's Generative AI Guidelines. The city has also helped convene a GovAI coalition to share playbooks with other local governments and, as reported by StateScoop, aims to expand the program to about 2,500 staff, roughly 30 percent of its workforce, by June 2027. Officials say the twin goals are to keep innovation in-house while creating clear, transparent oversight for any automation that touches residents’ data.
Other municipalities are experimenting with similar efforts, and San José’s mix of training, procurement controls, and an explicit push to have employees build their own tools has drawn national attention, as noted by Governing. Observers say the next big test will be whether the program’s early productivity gains can scale up without creating new privacy or equity risks. City leaders, for their part, describe the whole effort as deliberately incremental and practical: teach staff useful skills, put guardrails around how they use them, and let the projects prove their value on the ground.
