The Foundation for Agentic AI in Radiology

Introducing PAIRsdk — a new framework uniting radiology through intelligent collaboration.

A new developer framework designed to unify how radiology systems, imaging platforms, and AI models communicate and act — through a shared, structured foundation.

What Is PAIRsdk

PAIRsdk is RADPAIR’s developer framework designed to accelerate the adoption of Agentic AI across radiology.


Built with the intention of being open sourced, it provides the foundation for creating intelligent, interoperable, and voice-first workflows that connect people, platforms, and data.

PAIRsdk brings together five core components:

MCP (Model Context Protocol) — the connection layer linking AI tools and clinical systems.

RDS (Report Data Schema) — the canonical structure for radiology report data.

RDT (Report Display Template) — a standardized way to render and format reports.

RIP (Report Interaction Policy) — defines editing, confirmation, and user-permission rules.

AEP (Actions & Events Protocol) — a universal vocabulary for AI actions and workflow events.

    • Interoperability Without Reinvention
      PAIRsdk complements existing standards like IDR and FHIR, adding the operational layer those frameworks don’t define.

    • Efficiency & Safety
      Built for deterministic, auditable AI actions and role-based permissions.

    • Future-Proof Design
      Structured JSON and modular components reduce complexity and avoid vendor lock-in.

  • RADPAIR’s intention is to open-source the standard once coalition governance is established.


    Working together with industry partners, the goal is to finalize governance and publish the open standard in 2026.

Join The Watlist

Be first to explore the framework, contribute feedback, and help shape the governance process.

Are You Coming to RSNA 2025

Experience PAIRsdk & PAIR 3.0 in person at RSNA 2025, Booth #4932.
Join us in Chicago to experience it firsthand.

Book a Demo