Targeted AI adoption can drive change, current and former officials say

Federal agencies need to be strategic and forward-thinking when it comes to adopting artificial intelligence tools to enhance their missions, current and former officials said on Thursday during the Nextgov/FCW and ATARC public health IT summit.

John Boerstler, the general manager of Federal at Granicus who served as chief experience officer for the Department of Veterans Affairs from February 2021 through September 2024, said the agency has successfully scaled its uses of AI tools to better serve retired servicemembers. He cautioned, however, that “we need to really be careful about the implementation and in managing expectations, both internally and externally on how we use it.”

VA’s 2025 AI use case inventory, which was released last month, listed 367 examples of the department looking to onboard AI tools — a significant increase over the 227 it reported in its 2024 inventory. Many of these use cases are for medical devices or for other technologies meant to augment clinical care for veterans, as well as the delivery of critical services. 

Boerstler said looking at how AI can enhance customer experience was key, noting that a focus on “design and qualitative research” has also helped VA better engage with veterans. He said that, while working at VA, the agency found it was sending transitioning servicemembers around 156 emails in the last three months of their military service, many sent in the middle of the night, telling them to enroll in healthcare or file for benefits. 

To streamline these communications, Boerstler said VA “redesigned the entire journey,” including “using AI and automation to then set that new course and not spam your customers and help them enroll.”

“Now they're enrolling in healthcare at a much higher rate,” he added. “So there's so much that we can continue to do, and you don't need the fanciest AI model on the marketplace.”

Like VA, the National Institutes of Health has also been scaling its use of AI tools. 

Nick Weber, acting director of NIH’s Office of Scientific Computing Services, said the agency has focused on discussing “challenges and responsible and ethical AI use” across the organization, including by creating a generative AI community of practice roughly two years ago that now includes around 2,000 people.

To support NIH’s administrative-focused operations, Weber said the agency has also been using AI to better analyze its grants and funding efforts. He noted that approximately 85% of the agency’s $50 billion annual budget goes toward funding research across the U.S., and that “using AI to analyze that portfolio ​​— see where there are gaps, see where there's overlaps, where we can bring things together more efficiently — has been incredibly helpful.”

When it comes to NIH’s direct research efforts, Weber said the agency is trying to find ways to take the data it generates and collects and then build “domain-specific small language models, not the big frontier foundational models that exist.”

This entails looking at creating models “associated with our core data sets for Alzheimer's disease,” for instance, and then making them available to the broader research community so researchers “can ask questions and work with things more specific to a disease domain,” he explained.

In his second term in office, President Donald Trump has also placed an emphasis on enhancing government services through the adoption of new technologies like AI.

One such initiative is the General Services Administration’s OneGov initiative, which was launched in April 2025 to offer agencies discounts on select private-sector technology and software services — including AI tools — by treating the government as one customer. More than a dozen companies have reached deals with GSA so far.

“We've taken advantage of the OneGov deals at [the Department of Health and Human Services] and NIH for many of the $1 deals with the various products,” Weber said, adding that, while the initiative is “a great way to partner with industry,” he has some concerns about “the drug dealer model” of the effort when the discounted rates sunset.

"They get you hooked, and then you'll pay anything to continue to use it,” Weber said. “And so we're worried about, what is it going to look like? How are we going to decide when we have people using all of these different products? Where do we invest? How do we invest to make sure people can continue to get the benefits of the products? And we've been kind of really thinking about, “how can we take advantage of additional OneGov deals?’”