The Department of Veterans Affairs has published its comprehensive 2025 AI Inventory, documenting 367 individual use cases and 13 consolidated categories across the organization. This initiative addresses the critical need for transparent governance as AI adoption expands to improve Veteran access to care, benefits, and services. The inventory highlights measurable results including a 22% reduction in mortality among high-risk opioid users, significant workforce productivity gains with over 95,000 VA GPT users saving 2–3 hours weekly, and enhanced cancer detection through AI-assisted colonoscopy. By standardizing how AI tools are tracked and evaluated for safety, fairness, and effectiveness, VA demonstrates how responsible AI governance can deliver both operational efficiency and better outcomes for Veterans.
Case Study Source: VA Artificial Intelligence
Problem Statement
VA needed a transparent, governed way to track and scale its growing use of AI across the department, improving Veteran access to care, benefits and services while ensuring systems remain safe, secure and responsible.
Goal
Publish a comprehensive 2025 AI Inventory that documents AI use across VA, strengthens responsible AI governance, and highlights measurable improvements in health care, benefits processing and operational efficiency.
Challenges
Coordinating and cataloguing AI activity across a vast department, covering 367 individual use cases and 13 consolidated categories.
Ensuring AI tools meet rigorous standards for safety, fairness and effectiveness while usage expands.
Maintaining transparency to enable collaboration and oversight across internal and external partners.
Reducing administrative burden and improving service quality at the same time.
Flagging potentially fraudulent direct deposit changes that occur in roughly 1–2 per 1,000 updates.
Actions
Published the VA 2025 AI Inventory with both detailed individual entries and a consolidated overview to improve transparency and governance.
Collected standardised information on purpose, outcomes, development stage, benefits, and risk considerations for each use case.
Highlighted widely used AI capabilities and tracked scale of adoption, including licence counts, to reveal enterprise trends.
Deployed operational AI solutions such as VA GPT for staff productivity and AI-assisted software development tools.
Implemented clinical AI, including STORM for opioid risk mitigation and FDA‑approved colonoscopy computer vision tools.
Built an AI model (Payment Redirect Fraud) to identify direct deposit changes likely to be fraudulent for investigator review.
Key Results
Impact
Veterans benefit from safer, higher‑quality care, with measurable improvements including a 22% mortality reduction and higher adenoma detection rates.
Staff experience tangible time savings each week, freeing capacity for higher‑value work and improving job satisfaction.
Stronger governance and cross‑VA coordination through transparent cataloguing of 367 use cases and 13 consolidated categories.
How the VA Built Transparency Into Its AI Programme
The US Department of Veterans Affairs faced a significant challenge: artificial intelligence was spreading rapidly across the organisation, but there was no central system to track it. Without proper oversight, the department risked losing sight of how these tools affected the veterans it serves. The solution was to create a comprehensive inventory that would bring every AI system into the light.
The Scale of the Task
Documenting AI use across such a large organisation was no small feat. The final inventory captured 367 distinct AI applications, grouped into 13 broader categories. Each entry needed standardised information about its purpose, stage of development, benefits, and potential risks. The team also had to ensure every tool met strict safety and fairness criteria whilst usage continued to grow.
One particularly tricky problem involved fraud detection. Direct deposit changes happen frequently, but only about 1–2 in every 1,000 are potentially fraudulent. Spotting these needles in a haystack required a sophisticated approach.
What They Actually Did
The VA published its 2025 AI Inventory with both detailed records and a high-level summary. This dual approach gave different stakeholders what they needed, whether deep technical insight or a strategic overview.
On the operational side, the department rolled out VA GPT to boost staff productivity. Developers received AI-assisted coding tools. Clinical teams gained access to STORM, a system designed to reduce opioid-related risks, and FDA-approved computer vision technology to improve colonoscopy screening. A custom fraud detection model now flags suspicious payment changes for human investigators to review.
The Results Tell the Story
The numbers demonstrate real impact. More than 95,000 staff now use VA GPT. Over 70% report feeling more satisfied at work, saving 2–3 hours each week on routine tasks. For developers, the time savings are even more dramatic: over 2,000 people using AI coding assistants save 8+ hours weekly on average. Three-quarters say this frees them to tackle more meaningful work.
The clinical results are particularly striking. STORM’s deployment correlates with a 22% drop in mortality amongst veterans at high risk from opioid use. Meanwhile, AI-enhanced colonoscopy screening improved adenoma detection odds by 21%, with detection rates rising by roughly 4% overall. These aren’t marginal gains—they represent lives saved and cancers caught earlier.
Why This Matters
The inventory itself became a governance tool. By cataloguing hundreds of systems in a consistent format, the VA created visibility across the entire department. Teams can now see what others are building, avoid duplication, and share lessons learned.
Veterans receive better care because clinical AI systems are delivering measurable improvements in safety and diagnostic accuracy. Staff spend less time on administrative drudgery and more on the work that drew them to public service in the first place. And leadership gains the transparency needed to make informed decisions about where to invest next.
This case demonstrates that responsible AI adoption at scale requires more than good intentions. It demands systematic cataloguing, rigorous safety standards, and a willingness to measure real outcomes. The VA’s approach offers a template for other large organisations grappling with similar challenges: document everything, standardise your approach, and let the data guide your decisions.
Case Study Source: VA Artificial Intelligence
These industry AI case studies featured on our site are based on publicly available sources and are presented for informational and educational purposes only; we do not claim ownership of these case studies or affiliation with the companies mentioned, and attribution is provided where applicable.