Sourcing

What the optimizer already does well.

We concede the number, completely. Spend by category and supplier, price variance, contract coverage, the cost-minimizing allocation. An optimization engine computes all of it, and on the arithmetic it wins. We do not compete there, and we do not hedge about it. Nexonomy is not here to recompute a saving the optimizer already nailed.

The question the number cannot answer.

Whether the saving is worth the single-source dependency it creates. When to switch from an incumbent. How to weigh resilience against price. An optimizer prices none of that the way a human owner does, and reasonable people disagree about where the line sits. The output is not an allocation. It is a call someone signs and accepts the exposure for.

An optimizer gives you a number. A sourcing call needs a rationale you can sign.

One call, captured.

Nexonomy captured that exact call as a typed, content-addressed Decision Record, sealed the moment it was saved: the option chosen, consolidate to a single source with a backup retained, and the alternatives rejected, the concentration risk priced and accepted on the record, the confidence and why, the sign-offs, and the hash. One record, there on the day it matters. It runs watch-only inside your own environment, on the models you already trust, and nothing leaves.

How the call gets made.

Behind that record is the part an optimizer does not have. Several frontier models advise the same call in parallel, each weighing the saving against the single-source dependency it creates. Where they agree, confidence rises. Where they disagree, the split is surfaced and kept, not averaged into one number, and the recommendation that reaches the signer carries the dissent with it.

Everything that went into the call.

• The accepted exposure

  The concentration risk priced and signed for, not left unpriced by the optimizer.

• The alternatives rejected

  Multi-source kept on the record as the option not taken.

• Resilience weighed against price

  The trade-off the number cannot make, made and recorded.

• Replayed when conditions change

  When the supplier's health slips, the call is re-examined under today's assumptions.

When the assumptions move.

A year on, the supplier's lead times slip and demand rises; the saving that was optimal last quarter is the exposure this quarter. The question lands: who accepted this risk, and on what basis? On a normal stack the optimizer that produced the number kept no rationale, the analyst who ran it has moved on, and there is nothing to reconstruct but the saving itself. The exposure it hid has no owner.

Nexonomy kept the rationale the optimizer never had. The call was reached by the model panel weighing the saving against the dependency and splitting on resilience, and that reasoning sealed as a typed, content-addressed object: the option chosen and the alternatives rejected, the model split with the dissent kept, the concentration risk priced and the exposure accepted, the confidence, and the sign-offs.

When conditions move, retrieval is a lookup, not a rebuild: the priced exposure, the rejected alternatives, and the sign-offs exactly as captured, addressed by the hash of their canonical content so any copy verifies against the original, and the call surfaced for re-examination against today's numbers. The defense is completeness: the trade-off was priced and signed when the call was made, and reconstructs exactly. An optimizer is commodity, and so is a hash. The part that re-fits every model cycle is the advisory that prices the trade-off the optimizer will not, kept behind a record format stable enough that a call from last year still replays under this year's conditions.