Ten years ago, Kimberly McManus was building and analyzing early language models that were largely the domain of highly technical specialists.
Today, artificial intelligence (AI) is accessible to nearly anyone – and Kimberly, now acting chief AI officer at the Department of Veterans Affairs (VA), has a chance to help turn that shift into better public services at enormous scale.
“This is my first big technological paradigm shift of this magnitude,” Kimberly told MeriTalk in a recent interview. “It’s been really exciting to be able to get to help shape that.”
For Kimberly, the opportunity is not simply to make VA a leader in AI. It is to make sure veterans feel the benefits of the technology quickly, even if they rarely see the AI itself.
“My number one goal is to ensure that veterans benefit from advances in AI as quickly as possible and in an trustworthy way,” she said. “Veterans and American citizens should just accept nothing less than state-of-the-art public services.”
AI results are opening new doors
VA has been using AI in clinical environments long before the current generative AI boom.
Kimberly pointed to the Stratification Tool for Opioid Risk Mitigation, or STORM, as one of VA’s established examples of clinical AI. The tool uses predictive analytics to identify veterans at elevated risk of opioid-related adverse events so care teams can intervene.
Implementation of the system has been associated with a 22% decrease in mortality.
For Kimberly, programs like STORM demonstrate that VA already has “clear pockets of excellence” in AI. The challenge now is spreading that experience across one of government’s largest workforces and applying the same rigor to a much broader set of workflows.
Ambient clinical documentation is one of the areas scaling quickly.
More than 12,000 providers are now using the technology, Kimberly said. The tools can record, transcribe, and summarize conversations with veterans, with clinicians reviewing and approving the resulting documentation before it enters the medical record.
The goal is straightforward: less time on paperwork, more time with patients.
“We’re seeing that that’s both saving time and also improving the quality of the information and documentation,” Kimberly said.
She is equally enthusiastic about what AI can do farther behind the scenes.
Teams are using AI to triage IT incidents and accelerate software modernization. Kimberly said one modernization effort moved more than seven times faster than initially expected.
“Things with a 7x speedup just open up so many doors to building better services and solving incidents faster,” she said.
VA is also applying AI to benefits protection. A model designed to flag potentially fraudulent direct-deposit changes sends suspicious cases to investigators for review – helping catch attempted theft before veterans lose their benefits.
Taken together, the use cases span health care, benefits, software delivery, IT operations, and employee productivity. That breadth is part of what makes VA unusual.
Kimberly said the department has the highest number of high-impact AI use cases across the federal government and is now working through what the next stage of AI – including AI agents – should look like.
Scaling AI means scaling the people around it
For Kimberly, broader adoption cannot come from simply giving VA’s more than 400,000 employees access to new technology.
VA is building communities of employees who learn the tools, develop useful applications, and then help their peers do the same.
That means finding early adopters, giving them room to experiment, and turning their successes into examples others can follow.
VA runs communities of practice, internal showcases, office hours, and a weekly newsletter with practical tips employees can immediately test. The department has also begun holding hands-on, in-person enablement sessions at VA medical centers.
Rather than giving clinicians a generic AI presentation, teams spend the day working alongside them on actual use cases.
The approach also gives VA visibility into what employees really need before the department proliferates duplicative tools.
Kimberly said the goal is to “make it easy to do the right things” – giving employees approved tools, clear guidance, and practical support.
That becomes especially important as VA moves beyond early innovators.
“We had the early innovators, we had the early adopters, and now we’re kind of entering the more general group of employees,” she said.
That next stage, she said, will require much more focus on workforce enablement, training, and redesigning workflows end to end rather than simply improving one step in an existing process.
The best AI may be the AI veterans never see
Asked to rank VA’s fiscal year (FY) 2027 technology priorities, Kimberly put digital government and citizen experience first.
Cybersecurity and risk management came second, followed by AI and machine learning and workforce transformation.
The order reflects how she sees the technology stack: AI is an increasingly powerful lever, but the mission is the experience VA ultimately delivers to veterans, their families, and caregivers.
“Everything else is in service of that,” Kimberly said.
That is also why she does not necessarily want veterans thinking about VA’s AI strategy.
“While my role is to lead AI at VA, I mostly want that to be relatively invisible to veterans,” she said.
What veterans should notice instead are contact centers that can answer questions when they need help, health care appointments that are easier to schedule, and benefits delivered faster and more accurately.
Her vision isn’t just to make VA easier for Veterans to navigate. AI creates an opportunity for VA to be much more proactive, recognize what a Veteran may need, connect the right information, and bring the next step to them.
“We want technology to be amazing, but behind the scenes and in service of that experience,” Kimberly said.
Trust must scale alongside adoption
Kimberly’s second-ranked FY27 priority – cybersecurity and risk management – is inseparable from that goal.
“Veteran trust is really the foundation of our AI strategy,” she said.
That includes security, privacy, transparency, and maintaining human oversight in high-impact uses of AI.
VA’s data infrastructure is another major part of the equation. Kimberly’s team is working to make data easier to find and access appropriately, with the connectors and governance needed to put it to work without sacrificing privacy or zero trust principles.
She sees a major opportunity in connecting AI tools more directly with VA data systems so employees can complete more work in one environment rather than moving among disconnected systems.
At the same time, AI is raising the stakes for cybersecurity itself as models become more capable of finding and exploiting software vulnerabilities.
“Any AI tool we use, any AI, goes through our same rigorous privacy and security and authority to operate type processes,” Kimberly said.
Agencies should start small and quickly scale what works
If Kimberly could change one thing about how government modernizes, she would make it easier to shift resources quickly toward what is working – and away from what is not.
“If we could follow the venture funding within government, that would be amazing,” Kimberly said. “Start small, see if it works. If it does, figure out how to scale it, both technology-wise and funding-wise.”
And if it does not work?
“Scale it down, and that’s okay too,” she said.
Kimberly sees a similar need for technical talent inside government – people who can distinguish a promising technology from a useful one and help agencies deploy it responsibly at scale.
“There’s just no place more exciting and more impactful to work on than AI in VA,” she said.
The combination of scale and mission is difficult to match. So is the opportunity in front of the department.
For Kimberly, moving responsibly does not mean making veterans wait while government watches the technology develop somewhere else.
The longer-term opportunity is a VA that can anticipate needs, offer care and benefits proactively, and complete more of the administrative work behind the scenes.
“Veterans should not have to wait for the future,” she said. “They and their caregivers should be able to benefit from advances in AI as quickly as anyone else in America.”
More from MeriTalk’s Federal CXO Spotlight series:
- Basia Sall, Chief Data Officer, Nuclear Regulatory Commission
- Adam Starr, Chief Information Officer, Office of Personnel Management
- Mangala Kuppa, Chief Information Officer, Department of Labor
- Dawn Zimmer, Chief Information Officer, Department of Energy