Two-sided journeys · Evidence design
Role Fit
Making the evidence behind a job match inspectable.
Live in three languages, with source-linked briefs and Agent follow-up.
01The project
A live two-sided decision product that helps candidates organise fit evidence and helps recruiters verify the basis of a match, then improves the flow through testing and user feedback.
What this case provesDesigning a complete two-sided user flow and converting testing feedback into product iteration
02The problem
Candidates need to compare a job description with scattered experience evidence and identify honest gaps; recruiters need to verify why a person may fit without reading an entire portfolio or accepting a black-box score. The two sides share evidence but make different decisions.
03Key decisions
The product accepts pasted or uploaded job descriptions, retrieves verifiable evidence from the portfolio, and produces strong matches, transferable experience, honest gaps, and follow-up questions. Source cards keep every conclusion inspectable; candidates use the brief to prepare, while recruiters use the same evidence to continue their own assessment.
04Evaluation
In the 26–28 August 2026 invitation-only beta, 10 participants completed two feedback rounds and 8/10 adopted revisions. A matched-material regression fell to 0/5 normal reports; after repair, all 5 obtained normal reports, including one case that required operator-assisted retry recovery. That case is tracked separately from unaided completion. Adoption reflects feedback 30 minutes after report freezing, not long-term behavior.
05Outcome and current stage
The public matching page is live in three languages with text and file input, source-linked evidence briefs, Agent follow-up, and explicit next actions. An invitation-only candidate beta extends the same product logic through consented materials, reports, and feedback without changing the public flow.
Technical details & evidence
Role · Sole product owner and builder — discovery, product strategy, interaction, evaluation, launch, and iteration
- Product Discovery
- Evidence Retrieval
- Human-in-the-loop