As Centers for Medicare and Medicaid Services matures its uses of AI, the agency’s top IT official said monitoring how the technology contributes to outcomes is taking precedence over usage metrics alone.
The comments from CMS Chief Information Officer Patrick Newbold came during a panel at GovCIO’s Health IT Summit in Bethesda, Md., on Wednesday. Newbold said that while there was a focus on how many people were using the technology early on, that’s now shifted to an approach that factors in what the agency is delivering with its technology deployments.
“Instead of leading with AI, we’re going back and we’re leading with outcomes,” Newbold said while speaking on a panel about AI foundations in federal health. “And if it just so happens that AI is helping meet those outcomes, that’s a good thing.”
For months, CMS has touted that roughly 80% of its workforce is using AI, including its flagship AI chatbot known as “CMS Chat.” Also at CMS employees’ disposal: Over 70 use cases reported in the Department of Health and Human Services’ 2025 AI use case inventory in addition to major generative AI tools that were turned on for the entire department through discounted deals as part of the General Services Administration’s OneGov initiative. But Newbold said what the usage metrics don’t account for is whether the technology is making improvements for the populations the agency serves. As part of that outcomes-based focus, Newbold pointed to a forum inside the agency that’s being used for that purpose.
According to Newbold, CMS leverages a forum it calls Tech Pulse to track and monitor outcomes for technology. That group began roughly 20 months ago and it’s modeled after a similar forum at the Social Security Administration.
In an interview with FedScoop on the sidelines of the summit, Newbold said establishing that group “was an opportunity to have a performance management-based forum to keep in track and push a lot of the key outcomes and objectives that we had from an IT perspective at CMS.”
Because that forum is outcome-driven, its metrics are objectives and key results. Newbold said an example of one of the items it’s monitoring is the agency’s goal to get to a multi-cloud environment by the end of this fiscal year. In that method of tracking, the tools and other capabilities used to get to the end goal “are enablers of the outcome,” he said. “That’s how it’s different.”
Newbold also said that usage is still a good metric to track, but it isn’t the primary thing the agency is interested in.
CMS’s approach comes as the landscape for federal AI use is evolving. The deals GSA struck for deeply discounted AI systems are expected to end soon, leaving many questions about what the bill will end up looking like for agencies.
When asked whether the agency is tracking tokens or keeping uses streamlined as part of its management, Newbold said CMS has applicable governance built in — but he also has “concerns.”
“There’s not a lot of information … coming from the foundational models on what the costs are going to be long term. We took advantage of some of the OneGov deals, but we know those are not going to be forever,” Newbold said.
As a result, he said, CMS has to bake in the right governance and “to make sure that we have a good understanding of the use and can control the use, so the costs don’t spiral.” He said the agency currently has the capabilities to support that.