Evidence & verification
Prosli treats interview evidence as immutable and weighs every answer for authenticity. This is the moat that keeps confident-but-shallow (or AI-assisted) answers from beating real ability.
The evidence ledger
Every recommendation is tied to a ledger of the exact claims and answers behind it. When you open a candidate, you can trace any verdict back to the moment in the interview that produced it. There is no black box to argue with.
Answer authenticity
Prosli watches for responses that may not be the candidate's own — for example, answers that read as pasted or AI-generated rather than reasoned in the moment. When authenticity is in doubt, the answer is flagged and held for a quick live check instead of being taken at face value, and the candidate's confidence is lowered accordingly.
An authenticity flag is a prompt to verify in a short live round — never an accusation, and never grounds on its own to reject someone. A candidate may legitimately paste from their own work; Prosli says so when that is the likelier reading.
Immutable evidence
Prosli never edits, regenerates, or overwrites a transcript, an answer, a verification finding, or a recommendation. Re-running an evaluation with the latest engine is allowed; rewriting history is not. This is what makes a Prosli decision defensible weeks later.
Confidence and uncertainty
Every assessment carries a confidence read based on how much evidence was gathered and how well it held together. Uncertainty is always shown, never hidden — if Prosli could not verify something, it tells you what to check.