-
-
-
landing page
-
ai voice call
-
dashboard
Inspiration
1 in 5 hospital patients are readmitted within 30 days of leaving the hospital, and Medicare has been fining hospitals for it since 2012. The fix is obvious: just call patients after they leave. After discharge, someone still needs to Identify patients who need follow-up.
What it does
Once a patient is discharged - the AI calls with a follow-up call on day 1, day 3, and day 7, asking structured questions shaped to their specific procedure. A knee replacement patient gets asked different questions than a cardiac patient. If we hear any red flags—like a patient mentioning chest tightness, scoring their pain an 8, or saying they've stopped their medication—that's where we bring in the professionals for an immediate callback. For everyone else, it’s a smooth and reassuring check-in, all neatly logged in their record, without using up any staff time. Every call syncs to the hospital's software - giving clinical teams a complete post-discharge picture.
How we built it
Created Backend Agent in google Cloud run
- Call scheduling agent: - . Used Gemini, and the tools available from the Call-E MCP server
- end call transcription gemini and call-e, conversation agent
Challenges we ran into
enrolling patients in a follow-up sequence. Dockerization of agents on cloud run was tricky .coordination of real time response to app layer involved managing agents get run call . Turning discharge int follow-up calls.
Accomplishments that we're proud of
A working prototype with a full frontend. backend deployed on cloud run deploy.
What we learned
Building taught us that healthcare AI lives or dies on followups.
