Why every AI score needs a receipt.
Black-box matching is easy to ship and hard to defend. How we made reasoning a first-class product surface — and why “trust us” is not a compliance strategy.
The problem with a number
A match score without an explanation is a claim, not a product. Hiring teams can ignore it, over-trust it, or worse — use it in ways they cannot later justify to a candidate, a works council or a regulator.
Show the work
In Onbordo, every score arrives with quotable evidence: which requirements matched, which gaps mattered, and what the model did not consider. The UI treats reasoning as a peer of the number, not a tooltip.
Humans still decide
Architecture matters here. The product can summarise, score and rank — it cannot auto-reject. That line is not a marketing line. It is how we keep AI useful without outsourcing judgment.
What this unlocks
Reviewers move faster because they trust the summary. Compliance teams approve faster because the audit trail already exists. Candidates get a process that can actually be explained.