The problem
The client had raised a Series A on the promise of a mobile-first health platform but had no app. Their engineering team was four people, all backend. The runway pressure was real and the App Store review queue was the longest pole.
They needed someone who could own the mobile build end to end: design polish, native behavior on both platforms, AI features that worked, and the legal scaffolding around healthcare data.
What we built
A Flutter app shipped to both iOS and Android in eight weeks.
- Symptom triage flow: A guided, AI-assisted intake that asked branching questions based on the user’s responses. Backed by Claude on Cloud Functions, with deterministic safety rails for any answer that touched red-flag symptoms.
- Care team chat: End-to-end encrypted messaging with the client’s clinical team, with AI-generated suggested replies for staff (never auto-sent).
- Records & sharing: Document upload, OCR, structured data extraction, and shareable reports for primary-care handoff.
- HIPAA-aware infrastructure: Firebase configured for BAA coverage, audit logging in Cloud Logging, secrets in Secret Manager, no PHI in client analytics.
How we approached it
A two-day kickoff to lock the MVP scope. We cut every feature that wasn’t required for the launch story and put it in a “v1.1” doc. The clinical team gave us a non-negotiable list of safety behaviors that became the test suite.
Weekly Friday demos with founders, clinical lead, and design. Apple’s review queue was the schedule risk: we submitted a working build at week four to start the review clock before the app was technically complete.
The outcome
- Shipped to App Store and Play Store on day 56
- 10,400 installs in the first month, 67% organic
- 4.7-star average rating across both platforms
- Zero PHI incidents in the first six months
- The clinical team adopted the AI suggested replies in 71% of conversations
The client’s mobile capability is now in-house. We handed off the codebase, CI, and runbooks at week ten.