Building an AI assistant for secure patient triage
We developed a HIPAA-compliant AI assistant that handles patient intake, triages symptoms, and routes to appropriate care, reducing wait times and freeing clinical staff.
Where they started.
HealthTech Solutions operates 40 urgent care clinics. Patient intake was a bottleneck: nurses manually collected symptoms, medical history, and chief complaints before clinicians could see patients. During peak flu season, wait times exceeded 90 minutes. They needed a solution that could handle intake safely, maintain HIPAA compliance, and integrate with their EHR, without adding clinical risk.
How we solved it.
We designed a HIPAA-first architecture: AI runs in their AWS VPC, no patient data leaves their environment, and we implemented BAA-compliant data handling from day one.
We fine-tuned a model on de-identified triage protocols and symptom-to-pathway mappings, ensuring the assistant never diagnoses, only collects, categorizes, and routes.
We built guardrails: no medication recommendations, no treatment advice, and output filters to catch any accidental PHI. The assistant is explicitly scoped as an intake tool.
We integrated with Epic via FHIR for demographic sync and with their scheduling system for automated appointment routing based on triage outcome.
We ran a 4-week pilot at 3 clinics with clinician oversight before rolling out to all 40 locations.
The outcome.
Intake time reduced from 12 minutes to under 1 minute per patient, a 12x efficiency gain for nursing staff.
99.5% HIPAA compliance audit pass rate with zero findings in the initial deployment review.
Average wait times dropped from 90 minutes to 35 minutes during peak season.
Clinical staff report higher satisfaction; they spend more time with patients and less on paperwork.
Let's create your success story.
Every case study started with a conversation. Tell us about your challenge and we'll show you what's possible.