Skip to main content

Leveraging AI to summarize patient records and support Veteran care

Product management

We helped VA explore how generative AI could support clinical staff efficiency. Through comprehensive product management, we transformed an undefined concept into a working MVP—pioneering VA’s first integration of Large Language Models (LLMs) with their Electronic Health Record (EHR) system.

Client

U.S. Department of Veterans Affairs

The challenge

VA physicians, nurses, and clinical staff spend significant time reviewing patient records before appointment, including lab results, patient notes, military history, and other information scattered across the VA’s EHR system.

VA was seeking to use AI to improve efficiency and reduce clinician burnout, but faced significant barriers, like:

  • No precedent for successfully integrating LLM technology with EHR systems
  • Physicians using LLMs must constantly switch systems and copy-paste data
  • Complex technical architecture and approval processes for VistA EHR access
  • Undefined product concept needing strategic direction

Our approach

Prototyping and product strategy

  • Developed working prototype with mock patient data using Replit connected to OpenAI API
  • Facilitated strategic pivot from general-purpose tool to clinician-focused application
  • Led discovery sessions to identify meaningful AI applications for clinical staff

Product management and planning

  • Created comprehensive product strategy, including requirements document (PRD) covering problem statement, solution narrative, target audience, success metrics, timelines, and risks
  • Built three-phase strategic roadmap from discovery through MVP deployment
  • Managed agile development coordinating across development, design, user research, and data science

Technical integration and innovation

  • Guided development of VA’s first Model Context Protocol (MCP) server
  • Created documentation explaining new technology to VA stakeholders
  • Pioneered LLM integration with patient notes and medical records

Keys to success

Blue Tiger provided the product leadership necessary to move forward this effort led by Coforma.

  • Documentation and knowledge transfer for emerging technology
  • Strategic clarity in an undefined problem space
  • Rapid prototyping to build stakeholder alignment
  • Cross-functional coordination across technical and clinical teams
  • Navigation of complex organizational and technical constraints

“Thank you for helping us carve a path through a lot of uncertainty.”
— Government Partner

Results and impact

In 4 months, we successfully delivered an MVP in a staging environment, establishing a foundation for AI-enabled clinical tools at VA.

  • Transformed undefined concept into clear product strategy addressing real clinical needs
  • Built VA’s first Model Context Protocol server connecting to VistA EHR, enabling secure LLM access to patient notes and medical records
  • Delivered MVP demonstrating feasibility of LLM-assisted patient record summarization
  • Created reusable infrastructure supporting future AI applications at VA
  • Established product roadmap and agile processes for continued development