01 Discovery
Customer interviews + segmentation.
We talked to hybrid athletes, mapped the unmet need, named four segments.
FormBase
Labs.
From ideation to a launched iOS app. We owned the whole arc.
Plus: Shipped a Claude-compatible MCP server on top of the data.
What we did
Idea to App Store, owned end-to-end.
02 User stories
Stories before screens.
Job-to-be-done for each segment, then the screen that does that job.
03 Engineering
Native iOS, built to ship.
Training engine, run analysis, sync. Production code from prototype on.
04 Roadmap
Roadmap to App Store.
Sequenced scope. Beta cohort. Launch. Then the next bet.
Integrations · Engineering
Real run data, in.
FormBase grounds its insights in actual runs, not promises. We wired Strava as a first-class input: heart-rate zones anchored to history, race goals set against measured pace, settings the athlete owns.
- OAuth + first sync on connect
- HR zones, race goals, units, injuries
- Disconnect at any time, no lock-in
Plus · Agentic AI
A Claude-compatible MCP server, shipped on top.
FormBase data, exposed as tools. Claude reads training history, suggests next workouts, answers in the athlete's voice. The product gets sharper. The data does more.
- MCP server in TypeScript, deployable in minutes
- Read tools for sessions, runs, programs, zones
- Write tools gated behind explicit user consent
Product strategy
We didn't ship features. We built a loop.
Every athlete carries two questions: what do I do today? and am I actually getting better? Most apps answer one and bury the other. We wired them into a single feedback loop — today's work becomes tomorrow's proof, and that proof pulls the athlete back to the work.
The one screen that says: do this, now. No planning tax, no paralysis — just the next session.
Every session charted — strength score, volume, the body changing. The proof that makes tomorrow non-negotiable.
One engine, two jobs: track the work, motivate the next rep. That loop — not any single screen — is the product.