Health, AI prototyping · Contract project
Rapid AI prototyping a health app, bridging Figma and Claude. I turned a family member’s scattered medical notes into a calm daily dashboard, going from a rough idea to a working app in 3 days.
Role
Research, product design, requirements, AI rapid prototyping
Tools
FigJam AI, Figma, Figma Make, Claude
Outcome
Idea to working app in 3 days
Challenge
A family member was managing a complex medication and wellness routine in scattered phone notes.
Their doctors couldn’t see a clear timeline of when a medication or vitamin was started or stopped.
Solution
A calm health log that trades passive tracking for clarity, built on three pillars:
01
Smart reminders
Reminders set to exact days and times, with quick presets like Every day, Weekdays and Weekends.
02
Streamlined symptom tracking
A frictionless logging mechanic that captures physical symptoms and daily mood metrics in under three seconds.
03
Lifecycle logging and history
One timeline of past data. Medications can move to a “Stopped” history or be deleted for good.
Framing
I mapped the who, what, where, when, how and why in FigJam, then used FigJam AI to condense the board into a clear problem summary.
That summary became Claude’s starting context. I fed it raw medical notes to stress-test edge cases like overlapping doses, so mood logs line up correctly with the medication timeline.
Design
In Figma I sweated the details: type, color, icons and the overall design language. I stripped out the gradients and extra icons the AI drafts kept adding, and tightened spacing so the app reads calmly.
Real routines don’t fit “Morning, Afternoon, Evening”, so reminders take exact times, multiple doses and specific days. Any medication can be stopped, which keeps it in the history, or deleted. The “Track meds” screen swapped big pill counts for small pill graphics you can read at a glance, even on a bad day.
Build
Figma AI agents set up the component properties and variants, so I could stay focused on flows and hierarchy.
With components and tokens in place, I moved to Claude and had a working prototype within hours. I acted as the quality check: fixing spacing, matching colors and smoothing the microcopy. Last, I added an agent that answers a doctor’s quick questions using only the logged history, with no guessing.
Outcome
Handing the repetitive work to AI left more time for design craft. Next, I’m working with an engineering partner on native iOS and Android builds, a closed beta with people managing complex routines, and an App Store release.
Try it: the working prototype, right here. Tap through onboarding, add a medication, log symptoms or ask the agent.
Next project