Utah technology leaders are recommending statewide adoption of Claude Code for software developers and expanding the state's AI partnership with Anthropic after a pilot exceeded expectations.

The details: In early 2026, Utah's Division of Technology Services ran a four-week Claude Code pilot with 34 developers across 10 executive-branch agencies.

Alan Fuller, Utah's chief information officer, and Christian Napier, Utah's director of AI, discuss what the pilot revealed and what comes next for AI across Utah's public sector.

1. Why now: Why did you start experimenting with AI through the Claude Code pilot?

Fuller: We've been looking at AI for development tools for quite a while, and I really feel like when Claude Code launched in early 2025, there was a big step forward for helping to give developers more power.

It seemed like the coding agents have improved significantly, making them much more useful. So, it felt like the timing was right.

• Initially, even some of our own AI team was skeptical of coding agents, but that has now changed significantly, and we're now really embracing these tools.

2. Key numbers: How did Claude Code make the biggest difference during the pilot? Were there any results you could share?

Napier: We took a very disciplined approach to our pilot. Right after the project launch, we did a survey to see how people felt just getting their hands on the tool. I was surprised at the results.

• 80% of participants were able to stand up a functioning working environment within 30 minutes, which to me is like "wow." Normally, it takes a long time to get tools integrated in an environment and get people starting to use them.

• 77% had what they called a "magic moment," or a win, within an hour. They were able to get up and running right away and do something impactful within 60 minutes.

At the end of the pilot, we asked developers how much time they thought using Claude Code saved them.

• 71% of developers said that over that four-week span, using Claude Code saved them at least 20 hours.
• About 29% said they realized more than 40 hours of savings.
• That's a 25% time-reclaimed metric, which I thought was really surprising.

3. The highlights: What was the biggest surprise coming out of the pilot?

Napier: Number one was how quickly people were able to get up to speed.

With a lot of technologies, you need a lot of time to become proficient in using the tools and get to a point where you can say, "Hey, this is actually positively impacting the work that I'm doing."

Another important point is that we saw an evolution of the role of the developer. Almost 9 in 10 engineers saw their roles shifting where they could think about things more strategically.

• They saw themselves more as directors or orchestrators as opposed to task executors.

One of the big concerns around AI is, "Hey, AI can take my job," or "my skills might atrophy if I'm using AI for everything." As an engineer, you might be concerned about your own potential diminished value and your ability to contribute.

• But what we saw is that by using these tools, people actually expressed a higher level of confidence in their own abilities to handle more complex projects.
• I wasn't necessarily expecting that going into the pilot, but for me it was a happy surprise.

Not only did the engineers see their role evolving, but they also saw the work they were doing as more satisfying than it was when they started the project.

That helped assuage at least some of the concerns that people have about the use of AI and the impact on the work they're doing.

4. The challenge: AI has been a big talking point in Utah. How did you approach this partnership with that in mind?

Fuller: We feel very positive and bullish about AI, even though we see a number of risks. We're very conscious of those risks and are working hard to mitigate them.

We approached the partnership with an effort to really focus on business outcomes. Our goal is to achieve a better customer experience, and we believe we can use AI to do that.

• The goal is not so much using AI for the sake of using AI. It's more about the business outcome that we can achieve using AI — doing things and making enhancements cheaper and faster, while better meeting the needs of our customers, citizens, and residents.

We're going to manage risks with AI policy and AI awareness training. Christian and the team have put together an AI factory where, as projects come up in agencies, we try to systematize the process for bringing those to production.

• We're doing risk assessments. We work closely with our data privacy and data security offices.

So, we're trying to focus on driving the positives, while not losing sight of the potentially negative impacts and making sure we're addressing those as well.

5. Looking ahead: How is Utah thinking about the evolving role of AI in state government?

Fuller: I feel like it's still early days with this technology, and we see lots of use cases where we hope we can use it effectively.

We see a lot of opportunities in chatbots. But fundamentally, we want to transform the way the state does business — not just make an existing process more efficient, but come up with a new and better process using AI.

We also see a lot of opportunities with agentic AI, but we will need to have good governance.

• We're going to need to change some of our business processes, and I don't think we're anywhere close to seeing the full potential of these tools yet. We've got a lot of room in front of us to develop that.
• The AI landscape is changing very quickly, so it's hard to predict how we'll be using it in six months or a year.

In society, the use of AI is already becoming mainstream. We had a recent newspaper poll saying 80% of Utah adults are using AI regularly.

• But it's so early in the process that I don't think we're even close to seeing all the different ways it's going to impact society.

Hopefully for us at the state, we'll see an increase in efficiency. We want to improve the quality of service delivery. We need to be careful about ensuring compliance, and hopefully AI will give us better ways and better data to help our people — our humans — make more informed decisions.

We see tremendous potential. We see a lot of risks. And we're really hoping to be successful with the technology.

Learn more about Claude Code.