The State of Utah's relationship with Google Workspace began in 2012 during a migration from Novell GroupWise. Unlike the legacy system that offered limited functionality, Google Workspace provided a comprehensive productivity suite spanning Gmail, Google Calendar, Chat, Docs, Drive, Forms, Meet, Sheets, Slides, and more.

"The whole collaboration suite of Google Workspace has really improved our ability to support core initiatives like our rural work program," says another state employee. "Google Meet is one of our most powerful time-saving tools because it gives us the ability to attend meetings from anywhere in the state."

The cloud-native approach transformed how state agencies work. Multiple employees can now collaborate in real time on documents and presentations, eliminating the constant back and forth of emailed files. For example, when one employee recently had to give a presentation to the Department of Health and Human Services, they didn't have to send a file to anyone. They just asked everyone to refresh the Sheets link to see the most recent version.

When generative AI emerged, Utah's technology leadership saw an opportunity for further modernization. They also recognized the need to validate its impact before committing taxpayer dollars to a large-scale rollout.

In June 2024, the State of Utah launched a carefully structured pilot program, distributing 257 Gemini licenses across diverse state agencies. Participants had to complete mandatory responsible AI training and agree to abide by the state's generative AI policy. The team organized kickoff meetings, established bi-weekly training sessions, created a dedicated Chat space for sharing use cases, and conducted regular impact surveys.

The goal: to validate whether generative AI would actually save people time and improve the quality of their work. The results exceeded expectations. High-volume users — those interacting with Gemini 40 or more times in a 28-day period — reported saving an average of three and a half hours per week.

Then Google Workspace announced that Gemini would be integrated into the platform's core offering for all users. "Having Gemini included in all our Google Workspace licenses changed everything," the Division of Technology Services employee says. "We can now offer a generative AI solution to everybody in the state."

The pilot then pivoted to production. Working closely with Google Workspace engineers and implementation partner CVE Technologies Group, the State of Utah team navigated privacy impact assessments, security reviews, and a thoughtful opt-in process that respected each agency's readiness and needs. By February 2025, agencies began going live with Gemini, with most now having at least some employees actively using the tool.

The production rollout validated what the pilot suggested: When given the right tools and training, government employees will embrace AI to work more efficiently. In a July-August 2025 impact survey of active Gemini users, 27% of roughly 10,000 users responded — an impressive participation rate that reflects genuine engagement with the technology. The survey results showed that 100% of respondents saved at least one hour per week:

• 50% of users save an hour or more per week

• 27% save one to two hours weekly

• 16% save three to five hours weekly

• 7% save six or more hours weekly

"We're not forcing anyone to use Google Workspace with Gemini, so for me, the biggest surprise is the rate of adoption," the state says. "It's not often that we introduce a new technology and within a few months, close to half of the employees in the state are using it when they don't have to."

The use cases for AI at the State of Utah span the spectrum of government work. Some use Google Vids and NotebookLM to make content accessible and consumable for employees across the state, while others rely on Gemini to refine writing and transform bullet points into polished documents.

One employee even tested Gemini's extended context capabilities in an unconventional way: by asking Gemini to interview him from a list of 100 questions he created. "When I tried this experiment with other AI tools, they would get to maybe question nine or 10 and then go off the rails. Gemini remembered 99 of the 100 questions, so that was a real eye-opener to what this tool can do."