TL;DR
I co-led a 4-person team to design EMScribe360 — an offline-first, hands-free narrative assistant for EMS clinicians. Through 100+ stakeholder conversations and 60+ deep interviews and ride-alongs, we mapped the EMS workflow, defined an MVP, and produced a business evaluation with clear risks and next-step experiments.
North star: reduce narrative time and cognitive load without sacrificing accuracy in chaotic, high-noise environments.
Context
EMS clinicians must document every call in an electronic patient care report (ePCR). Structured fields — vitals, checkboxes — are manageable. The narrative is the hard part: it has to capture the full "story" with clinical and legal precision.
Reality: narratives are often written hours later, after multiple calls, restocking, and handoffs — when memory fades and fatigue is high. Opportunity: capture key details in the moment, structure them automatically, and export clean text into existing ePCR systems with no double documentation.
Problem
"The structured fields are fine — it's the narrative that drains you."
— Synthesized insight from EMT/paramedic interviews
- Delayed documentation → memory decay. Narratives written hours after the call create omissions and inconsistencies.
- Chaos and noise break generic voice tools. Background audio, interruptions, and glove constraints make standard dictation unreliable.
- Handoff is high-stakes and ephemeral. The first minute of ED handoff is critical; afterward, nuance is hard to recover.
- Interoperability is mandatory. Any tool must export cleanly into existing ePCR systems — otherwise it adds work instead of removing it.
My Role
- Discovery planning and synthesis (themes → requirements)
- EMS workflow mapping (job map) and MVP scope definition
- Product strategy artifacts (positioning, constraints, risk assumptions)
- Early UI direction and solo MVP build plan
Approach
1. Discovery at scale. Conducted 100+ stakeholder conversations and 60+ deep interviews and ride-along observations across EMT and paramedic roles.
2. Synthesis → Job Map. Mapped the end-to-end EMS workflow (scene → assessment → treatment → transport → handoff → report) to pinpoint where details are lost.
3. Requirements and constraints. Translated field realities into hard requirements: hands-free capture, offline-first operation, EMS terminology accuracy, anti-hallucination safeguards, and ePCR export.
4. Solution concept and MVP. Designed a push-to-talk experience that collects timestamped voice notes with event markers and generates a structured DCHART narrative with human-in-the-loop edits.
Key Non-Negotiables
- Offline-first operation (no cellular dependency in the field)
- Fast capture under gloves and multitasking conditions
- Direct export to ePCR — no double documentation
- High accuracy: hallucination is clinically and legally unacceptable
Impact
Even with the product paused, the project produced durable outcomes: validated the severity of narrative documentation burden, translated 60+ interviews into a buildable workflow MVP, and clarified adoption risks early — before any engineering investment.
What's Next
- Incident creation and event timeline logging
- Push-to-talk audio clips with event markers (Scene / Assessment / Treatment / Transport / Handoff)
- DCHART narrative generator with human-in-the-loop edits
- Copy-to-clipboard export for ePCR paste-in