VA eyes AI to improve software license management

The Department of Veterans Affairs entity tasked with centralizing software management across the agency is eyeing artificial intelligence to more swiftly drive cost savings by offloading licenses that aren’t being used or are redundant.

Jeff VanBemmel, executive director of end user operations within VA’s Office of Information and Technology, testified Monday that the department’s Enterprise Software Assessment Management program, created in January 2024, has made significant progress in gaining visibility into software usage across the VA.

As the program nears a full view of all the software the VA has bought and how it’s using it, it will look to use AI to rationalize what software licenses aren’t being effectively used, are redundant or could be renegotiated to generate savings.

VanBemmel told members of the House Veterans’ Affairs Subcommittee on Technology Modernization that though using AI for software asset management is a “relatively new investment” that likely won’t be fully fielded until later this year, “we do see an opportunity on the platform to introduce AI to help with that work.”

To date, the VA’s Enterprise Software Assessment Management program has generated $136 million in cost avoidance just by reforming its approach and acquiring new tools for managing licenses with the top 15 commercial software programs the department uses, VanBemmel said.

And the program is pushing far beyond that, hoping to get “full visibility” of the VA’s software environment in the near future, he added.

“I think we have very good visibility on more than 80%” of the department’s software licenses, VanBemmel said. “We can see almost all of the software today, and we have made some investments in visibility on the endpoint. So we can actually not just see the licenses assigned to this user, but we can actually tell if that user is using that license. And that’s a piece of software that we did not have even a year ago.”

Once that visibility is in place, the VA will need to combine it with acquisition data — and that’s where AI could play a role.

“AI will help us with the aggregation of large data sets and help us understand what’s in use” versus what’s been bought, VanBemmel explained.

By crunching through that data using AI in a smarter, quicker way, it will then be up to leaders in VA’s Office of Information and Technology to decide how to move forward with acquired licenses to drive economies of scale, particularly in areas where the department uses several similar tools from different vendors.

“I think the bigger issue that we have in terms of redundancy is not duplicative software, but in a software category, we have four or five titles that do the same types of things, just in a different way,” he said, referencing commercial video call tools like Zoom, Microsoft Teams, Cisco Webex and Google Meet as an example of the issue.

From there, VanBemmel said, “the question would be what is the best value for VA in terms of reducing those four or five titles to say one or two different options. That’s the work that we have ahead of us, even after we get the asset management system in place — that is a long-term effort to try to go through all of those titles.”