Introducing Agent Interop
Healthcare Agent Framework
Introducing Agent Interop: A Framework for Agent-Based Healthcare
Care Gaps Are Broken
As I shared recently in my podcast with Benji Graham and Bill Lakenan: care gaps are broken.
For years, our industry has been built around gap lists Excel sheets, PDFs, and faxes passed around back offices. Patients remain in the dark while staff spend endless hours trying to close loops.
But imagine if instead of back-office paperwork, the patient became the center of the loop. Imagine being immediately prompted by your provider’s system:
Eligibility is checked automatically against medical guidelines.
Recommendations are issued and accepted in real time.
Care Gap is Closed in both Provider and Payer systems if the criteria for medical guideline is satisfied fully
Scheduling is completed instantly without staff chasing faxes or providers calling payers.
No manual intervention. No wasted hours. Just context-aware expert agents communicating in real time on behalf of patients, providers, and payers.
Healthcare is already 1/3 of U.S. GDP and projected to keep rising. We cannot scale by adding more humans to the system. We need smarter, automated, agent-based interoperability.
Welcome to the Agent Interop Framework.
What Is Agent Interop?
Agent Interop is an open framework for healthcare agent-to-agent communication, inspired by the pioneering work of Josh Mandel and tested with colleagues like William Laolagi and Pawan Jindal.
It builds on the Language-First Interoperability concepts from the HL7 Connectathon, supporting both:
A2A (Agent-to-Agent) JSON-RPC + SSE
MCP (Model Context Protocol)
With Agent Interop, healthcare agents can:
Negotiate eligibility in natural language.
Exchange structured FHIR resources (e.g.,
QuestionnaireResponse,Bundle).Guide patients directly into action flows like screening or referral scheduling.
✦ Use Case #1: Breast Cancer Screening
Our first demonstration is a Breast Cancer Screening (BCS-E) agent workflow:
Applicant Agent (Patient-facing) ingests EHR/FHIR data, confirms demographics and past screenings.
Administrator Agent (Provider/Payer) checks eligibility against guideline logic.
Decision + recommendation artifact is produced instantly.
Scheduling is launched immediately via SMART Scheduling Links to find open mammogram slots and book them.
No care manager queues. No back-office delays. Just real-time action at the point of need.
✦ Functionality & API
The Agent Interop demo includes:
FHIR connector: ingest real patient data (capabilities,
$everything)Scenario registry: Currently supports BCS-E, (future: clinical trial matching, prior authorization, specialist referrals)
Agent APIs:
A2A:
message/send,message/stream,tasks/get,tasks/cancelMCP:
begin_chat_thread,send_message_to_chat_thread,check_replies
Scheduling integration: (TBD) fetch slots from SMART Scheduling Links publishers, deep-link to booking portals
Trace panel: auditable record of decisions, eligibility rationale, slot choices
Docs + endpoints: Agent Interop GitHub.
✦ What’s Next?
We’ve started with Breast Cancer Screening, but the Agent Interop framework is adaptable to many critical use cases:
ADHD identification in young adults
Clinical trial matching, including rare cancer cohorts
Prior authorization automation
Specialist referrals with instant scheduling
✦ Why This Matters
The promise of agent-based healthcare is real-time, guideline-driven, patient-centered action.
Patients: prompted to act, supported with clear eligibility and scheduling.
Providers: freed from chasing gaps lists, faxes, and prior auth calls.
Payers: see decisions logged with transparency and auditability.
Agent Interop proves that with A2A and MCP, we can move from slow, manual, back-office operations to fast, automated, patient-first care.
✦ Join Us
Explore the framework: Agent Interop on GitHub
Learn about the Language-First track: HL7 Connectathon
Read the docs: https://agent-inter-op.vercel.app/docs | https://agent-inter-op.vercel.app/redoc
Try the demo app: https://agent-inter-op.vercel.app/
Huge thanks to Josh Mandel for blazing the trail, William Laolagi for tag-teaming testing, and Pawan Jindal for stress-testing and pushing the ideas forward.
Healthcare won’t fix care gaps with more gap lists. It will fix them with agents that act.
Welcome to Agent Interop.



Discussion about this post