Commodity and packaging contracts are negotiated on cost structures - agreed shares for labor, logistics, base material and margin, with risk shared between both sides. Prices are supposed to move with the indices at every revision cycle. The demand we keep hearing: digitize the formulas, link the indices, and let the recalculation run itself.
Commodity and packaging contracts are not negotiated as a single number. They are negotiated as a cost structure: agreed shares for labor, logistics, base material and margin, with movement risk deliberately shared between buyer and supplier. When the underlying indices move, the price is supposed to move with them - at every revision cycle, by the formula both sides signed.
The ask is direct: get those formulas out of the documents and into a system, link the commodity indices they reference, and make the recalculation automatic. Not a renegotiation - the agreed mechanics, executed without a spreadsheet marathon.
Of everything in that conversation, this was the sharpest pain - named as a current struggle and a strategic priority in the same breath.
Heard from a senior procurement-solutions leader at a global consumer-goods manufacturer. Paraphrased, like everything on this map.
Every formula contract is the fossil record of its own negotiation. The cost breakdown differs contract by contract - a share defined as a percentage here, a base value with an adjustment clause there, a cap someone conceded in year two. There is no standard shape to extract; each document has to be read the way a contract manager reads it.
The indices are no tidier. They come from different providers, on different cadences, in different units - and each contract picks its own reference points and averaging windows.
Directional only, like everything on this map. These pools are large enough without decoration.
Every late or mis-recalculated repricing settles money on the wrong side of the agreement, in both directions - and nobody goes looking for the errors, because nothing looks wrong.
The agreed formula runs the same governed way every cycle, so the silent errors stop being written - and the wrong price stops becoming the baseline for the next cycle.
Category managers spend revision cycles on recalculation mechanics - spreadsheet campaigns run per cycle, per category, per region, against index feeds that never sit still.
The mechanics move to the system and the judgment stays with the people. Revision cycles stop being spreadsheet campaigns and go back to being about suppliers, categories and terms.
When a supplier challenges a price, reconstructing which index value fed which output means an argument conducted across old spreadsheets and older memories.
Which index, which value, which date, which output - on the record for every contract, every cycle. The answer becomes a lookup instead of an argument.
Two engines run in production today - Analytical Lab Reports and account Knowledge Twins. Everything else on this map is an extension on the same foundation.
This entry splits cleanly into two halves. The document half is the Intelligent Document Agent mechanism - the same structure-out-of-documents capability that runs behind Analytical Lab Reports, pointed at contract cost structures instead of test results. It reads each contract the way a contract manager would, and stores the formula as structure.
The recalculation half is deliberately boring. Once the formula is explicit and the indices are linked, the math is deterministic - executed by systems, identically, every cycle, with every input on the record. AI reasons about the document. It does not do the arithmetic.
See the same parsing mechanism in productionAll of it runs inside your infrastructure, within your boundary. Negotiated cost structures are some of the most commercially sensitive text you hold - they stay under your control, and so does your freedom of action: which models do the reading, where the workloads run, what the economics look like.
Back to the AI ROI MapThe document mechanism this entry depends on runs in production on analytical lab reports. Take the open dataset, run the same extraction in the AI tools you already use, and try to prove us wrong. No form, no gate.
On-prem. Your data never leaves your boundary.