Reconcile scores against supporting evidence

Packaged know-how that tells an agent how to do a job well.

You might say…

“Assembly can drift ratings away from the data — I need a pass that checks each score still matches what it's meant to be based on.”

What it does

Walks each scorecard rating and judges whether it is justified by the evidence cited in support, flagging any score that the evidence does not bear out or that contradicts the rubric.

Trigger: Use to catch ratings that moved during assembly, before the artefact leaves the firm.

I/O: Assembled scorecard with ratings + evidence references → list of reconciliation flags

Recognise the problem?

The primitives are the commodity part. The fastest next step is a conversation about composing them into something that works for you.

Start a conversation