day10 / our work.
uCTRL — an AI-Native Health Intelligence App
Mobile
Healthtech
AI gives every user the same health answers. It does not give them an aha moment.
/01
Client
uCTRL is a health-tech startup built around a single thesis: your health data is yours, and you should be able to own, understand, and act on it. The product is currently in MVP, with university students as the launch audience and a broader roadmap extending to chronic-condition patients and health-conscious adults.
The team is built by veterans of the AI space, including founders who previously built Siri (before the Apple acquisition), alongside practicing physicians from top clinical institutions. For an early-stage health app, that's a rare combination. Every product decision sits at the intersection of consumer UX and HIPAA compliance.
/02
Challenge
01
Health apps don’t lack AI — they lack AI that knows you.
Wearables, EHR portals, pharmacy apps, clinic notes: the data exists, but it lives in fragments, behind separate logins, in formats that resist combination.
02
Existing solutions are vertical and walled — each one stores something, none of them reason across the others.
The result is insight as a feature buried in a tab: something the user has to go looking for, rather than something that meets them.
/03
Solution
Every day, the app surfaces AI-generated insights that connect events across the user’s data: sleep patterns, medication timing, activity trends — signals that would be invisible if you were looking at each source separately.
The team calls the first one the “aha moment”: the instant a user sees the app reason across their own life and surface something they couldn’t have found themselves. Not a feature. A revelation. That’s what the product is built around.
In clinical settings: scan a QR code, and the doctor has the full health record before the clipboard comes out. No forms. No memory required.
Traditional health apps assume the user fills in fields and the app stores them. uCTRL works the other way: AI generates structure. The client’s backend — a four-service Python stack on AWS, with a stateless Claude model on Bedrock and a dedicated insights service — produces reasoning, summaries, and connections on top of the user’s encrypted records.
Our job was to make that reasoning land natively: to turn an AI response into a card the user can act on, a setting the user can flip, a destination the user can navigate into. AI as a feature of the app, not a tab attached to it.
Past events inform current insights. And the data stays in the user’s control — encrypted, HIPAA-aligned, and built so it cannot leak across accounts even at the storage layer.
How we shipped
/01
Across a surface that made that pace meaningful: HealthKit integration, real-time voice, encrypted health records, AI insights surfaced as native UX. Each piece non-trivial. Together, more than a standard mobile build. Here’s how:
/02
Every piece of work, technical or not, lives as a Linear ticket; every team member always has one in-progress. The discipline keeps scope, owner, and definition of done explicit, and removes the back-and-forth of “what are you working on?” from the day-to-day.
/03
The iOS app is built with Claude Code as the only AI tool in the loop — codebase exploration, refactors, type-narrowing, ticket-by-ticket implementation, all paired with the model. The full surface area of the app — AI insights integration, lifeline timeline, sharing flows, medication tracker, voice infrastructure, onboarding — is shipped by a single embedded engineer at a velocity that would normally require a team.
/04
New UI concepts are prototyped in Replit and pressure-tested with real users before they reach the iOS codebase — so UX choices get confirmed at lower cost than a traditional design-to-engineering handoff, and engineering effort goes only toward what’s already validated.
/04
Impact
We built the iOS surface that turns uCTRL’s AI layer from infrastructure into product — the thing users actually open the app for. Daily insights that connect what no single data source could, a conversational assistant grounded in the user’s own record, granular sharing, medication tracking, and a lifeline timeline that makes ownership feel real. Less time translating AI output into UX. More time shipping features users actually see.
01
Before
After
01
02
03
04
05
06
/05
Team snapshot
AI-Augmented Senior Pod
1 FTE
+
Traditional build would require
3-4 FTE
Building from zero or transforming existing operations? Time to claim your unfair advantage
