The Challenge
A Series-A healthtech startup was drowning in manual patient intake and triage. Their care coaches were burning out — handling 400+ daily queries with spreadsheets and disconnected tools. Response times exceeded 48 hours and NPS scores were plummeting.
They needed an AI system that could automate initial triage without compromising the compliance requirements of handling protected health information (PHI). Equally important: the team needed sustainable wellbeing practices to prevent further attrition.
Our Approach
We assembled a dedicated Studio Sprint pod — engineers, an AI strategist, a UX designer, and a wellness concierge — for a focused 6-week engagement:
- Week 1–2: Discovery workshops with clinicians and engineers. Mapped the existing triage flow and identified 12 automation opportunities.
- Week 3–4: Built the core NLP pipeline (Python, spaCy) for symptom extraction and priority classification. Deployed on AWS with end-to-end encryption.
- Week 5: Integrated the triage assistant into their existing React dashboard with real-time WebSocket updates.
- Week 6: QA, HIPAA compliance audit, and go-live. Parallel launch of weekly listening pods for the care coaching team.
The Results
Within the first 30 days post-launch:
- 32% reduction in average patient response time (48hrs → 33hrs)
- 68% of queries auto-triaged without human intervention
- Zero compliance incidents during the first quarter
- Care coach burnout scores dropped by 41% after introducing listening pods
Tech Stack
React · Node.js · Python (spaCy, scikit-learn) · AWS (ECS, RDS) · PostgreSQL · Figma · HIPAA-compliant infrastructure