An AI triage layer that reads symptoms, history, and policy to route patients to the right level of care — consistent across every clinic, and compliant by design.
Talk to our healthcare team
MedCore Health Systems runs a network of more than 200 primary and urgent-care clinics. Triage quality varied site to site — driven by who happened to pick up the phone — and that inconsistency showed up as avoidable ER referrals and missed urgent cases.
They wanted one consistent, defensible triage standard across the whole network, delivered through AI, without putting a single byte of PHI somewhere a regulator would object to.
Triage was only as good as the person answering the phone. Across 200+ clinics that meant wide variance: some patients were over-referred to the ER, others were under-triaged and slipped through.
Two patients with the same symptoms could get two different answers depending on which clinic they called.
We built a triage layer that reads the patient’s symptoms, history, and the clinic’s protocols, then recommends a level of care with the reasoning attached. Confidence gating keeps anything ambiguous with a human clinician.
PHI is scoped and redacted on the boundary, and every recommendation is logged and reproducible — so the same case always gets the same defensible answer, and compliance is a review of configuration rather than an archaeology dig.
Claude
LangGraph
Presidio
FastAPI
PostgreSQL
AWS
Kafka
Redis
DockerEverything in one polished PDF, ready to share with your team.
Big idea, small idea, or somewhere in between — we’d love to hear it.
