MindGym
AI-Powered Mental Performance for Jobseekers
Designing an intelligent coaching platform that helps jobseekers build confidence, identify blockers and prepare for high-stakes moments.
50+ user interviews revealed that job searches stall through behavioural failure, not skill gaps.
Job searches stall through behavioural failure, not skill gaps — and behaviour only changes once mindset does. That moved the brief from "help users do more" to "find where this user is actually losing, and change how they feel before they act."
From first meeting through to production handoff.
50+ interviews synthesised into three distinct archetypes.
In a team of 13, clarity about individual contribution matters.
One colour, one meaning. Teal appears exclusively where the user should act next. Amber for time-sensitive. Green for positive delta.
Session setup could have been a single form. Breaking it into 6 focused questions made drop-off rates fall significantly in internal testing.
The Insights page surfaces 'Morning sessions produce +2.1 higher lifts before 10 AM' — not a time-of-day chart. AI analyses and concludes.
7→4 alone feels clinical. The session summary combines the number with qualitative tags, Maya's verdict, and next actions. Three layers make the shift feel real.
Three things I would change — not because the work was bad, but because I've learned from it.
I went to high-fidelity too quickly. Some onboarding decisions would have been better validated with paper sketches and 5 quick user conversations first.
The research was clear — the most critical moments happen on mobile. I accepted the desktop-first constraint too easily.
The PRD had clear targets but I didn't connect my design decisions explicitly to those numbers throughout the process.
A complete, production-ready AI product shipped end-to-end.