Interop Blog

Yesterday I had the pleasure of joining Farzad Mostashari and the Aledade, Inc. team (Nicholas Gerner, Ashok Srinivasan, and Jonas Goldstein) to explore the application of Model Context Protocols (MCPs) in healthcare IT.

We want agentic AI systems to handle real work (e.g., kick off a prior authorization or summarize glucose management history).

I accidentally hoarded OpenAI o3 query credits this month (the psychology of scarcity in the ChatGPT "Plus" tier!).

A recent letter from the HIMSS Electronic Health Record (EHR) Association to ONC outlines recommendations ostensibly aimed at "Smart Deregulation in Health IT."

What if, instead of pre-specifying every data field and every possible workflow step, we could enable conversations between capable systems?

Clinical workflows are full of necessary steps that don't happen instantly.

Another day, another open protocol!

Last week, I discussed how LLM agents embedded in EHRs could transform complex workflows like prior authorization by using tools to interact conversationally, rather than relying solely on rigid structured data exchange.

Prior authorization imposes a barrier to timely care.

Prior authorization is friction.