The morel is specifically designed to survive manual review. What we have found — consistently — is that it is sometimes embedded in the supplier's own system. Something their own employees aren't aware of. A pricing rule. A substitution logic. An algorithm that runs automatically and leaves no fingerprints at the line item level.
What exposes it is not a sharper eye. It is a pipeline robust enough to hold thousands of line items in simultaneous comparison — maintaining every pattern from every previous invoice and quote — mapping it across time — and drawing conclusions that trace back to the origin. That is not a manual process. That is an engineered one.
But the pipeline doesn't move first. Human intuition does. Over years of this work we have developed a feel for what we are looking at. The first conversation tells us more than most people expect it to. We listen for what you believe is happening. Then we go find it — or find what's actually there instead.
This is what the pipeline sees. Supplier nodes. Invoice nodes. Line item nodes. Pattern edges connecting them across time. The morel appears as a cluster — a concentration of connections that shouldn't be there. In a manual review it's invisible. At this scale it's undeniable.
Simulated data for illustration. Node types, edge weights, and cluster patterns reflect real engagement structures with all identifying information removed.