What we do, and don't, today.
Stated plainly, so you can decide whether to trust the evidence.
No fabricated candidates or outcomes
Every mission, session, receipt, and decision you see inside Fydell is backed by a real row in our database, created by a real action someone took. We don't seed demo data that looks like real people.
Simulations are simulations
Simulations run in your browser against authored scenario materials, not a live production system. No real customer data or systems are touched during a session.
Evidence is rule-based, and we say so
Findings on a work receipt come from simple rules over the recorded event timeline (command runs, file saves, chat messages, timing). Each finding states its confidence and its limitations. We don't present this as a certified psychometric assessment.
Technical failures aren't charged
If a session breaks for reasons outside the candidate's control, and we can confirm it, it's flagged as a technical failure and never billed, and never counted against your pilot total.
You control your own evidence
Candidates decide who sees their results. Employers only see evidence from simulations they ran themselves. Nothing is broadcast to a public network.
What's still early
We're a founding pilot. The network view, richer graph, and outcome tracking are intentionally minimal right now. We'd rather ship something honest and small than something impressive and fake.