Why “pass” is not enough?

Monday September 28, 2026

A product shipped two years ago. A clause was marked compliant at the time. Now someone wants to know why: an accreditor doing a routine check, an insurer after a claim, a court following an accident.

The honest answer is rarely "we don't remember."

It's usually worse: a checked box, and nothing behind it.

It's not the file itself that gets judged. It's the reasoning behind every verdict inside it.

A technical file is the manufacturer's package: product description, labels, drawings, existing test reports, a draft declaration of conformity. Assembling it is the manufacturer's job.

The Conformity Assessment Body (CAB) job starts after that. Check the file against every relevant clause of the applicable standard, and decide, clause by clause, whether it's met, missing, or unclear.

That decision is what actually gets challenged later. Not the file itself, since it already existed before anyone opened it, but the verdict on each clause, and the reasoning behind it.

Where the report gets audited, in practice

Notified bodies & Labs are themselves audited by Accreditation Bodies and part of what gets verified is simple: can you justify your own decisions after the fact? A reviewer asked why a specific clause was judged compliant needs tangible elements to prove compliance.

If the reviewer who made that call left, or genuinely doesn't remember 2 years later, the report must answer on its own. When it can't, that's not an awkward silence. It's a finding against the CAB, not against the manufacturer.

A verdict with no reason attached isn't a pass. It's a decision nobody can defend yet.

The same situation may apply later, outside any routine audit. If a product fails in the field, the first question is: Show the product conformity report.

Where AI actually helps

Manually writing a justification for every clause takes time conformity engineers rarely have. When a clause fails, the justification comes naturally: there's a specific gap to point to. When it passes, writing why takes just as much effort, and that's usually where the shortcut happens: the box gets checked, the reasoning doesn't get written down.

Holding a clause, its supporting evidence, and the reason it was judged sufficient together, consistently, across hundreds of clauses, is exactly the kind of structured, repeatable work well suited to AI support.

That's the part we built Qleer around. Not replacing the judgment call. Making sure the reasoning behind it is captured the moment it's made, so it's still there the day someone, an accreditor, an insurer, a court, asks why.