AI procurement negotiation preparation that holds a price increase to what the costs support
Price increases held to what costs support, every number in the brief cited.
See one case, screen by screen ↓AI procurement negotiation preparation for category managers. Eight agents read a supplier's price-increase letter, pull the spend, contract and the supplier's accounts, rebuild the price from the contract's cost indices, check every number twice, bring benchmarks and alternatives, and play the supplier in a rehearsal. The result is a seven-card brief with every number cited; the category manager sets the targets and the head of sourcing or CPO approves them.
How to choose AI solutions for back-office and enterprise work →
Updated
Suppliers keep writing to ask for more: resin, energy, labour, freight, tariffs. Each letter comes with a short reply window, and a proper answer needs the spend history from the ERP, the contract's price clauses, the supplier's filed accounts, the cost indices since the last increase, comparable settlements and the qualified alternatives. Pulling that together takes a category manager a full day, and the result lives in one spreadsheet nobody else can check.
So the work is often skipped. The letter's chart of the worst week of an index is taken at face value, the contract's raw-material adjustment clause is never used, nobody remembers that the moulds are ours, and the increase applies on the effective date because the reply window passed. Targets are never written down, so the outcome cannot be judged, and savings claimed after the meeting never reach the budget.
Estimated split for one major supplier negotiation prepared to standard — about a full day by hand, about an hour of the category manager's time with the solution.
Eight agents research, rebuild the price, check each other and play the supplier; the category manager sets the targets.
On 22 Sep 2026 Brantmoor Polymers, which moulds the polypropylene housings, grilles and covers for Kestrel & Vane Group's appliance plants, writes asking for 11.0% more on all 38 parts from 1 Jan 2027: €1,562k a year on €14.2M of spend. The reply is due Thu 22 Oct, the day of the meeting in Plzeň. Here is how Annika Brandt, category manager for plastics and packaging, prepares it with the agents, screen by screen, in the working solution.
Annika owns five open negotiations. Her day opens with a note from the brief writer: overnight the agents refreshed every source behind her briefs, and no number moved. Each item has one button. Brantmoor's brief v3 is ready and needs one decision from her; Pelham Industrial Gases asks 10.0% where energy, the only driver it names, supports 1.2%.
“Brantmoor Polymers asks 11.0% on €14.2M; the costs support 1.9%. Brief v3 is ready. The fact checker needs one decision from you: the labour figure.”
Rafael triages the portfolio: 23 open letters, €8.06M a year claimed on €115.6M of spend, and about €3.56M of it supported by costs. Each row shows what was asked, what the costs support, the reply date, the owner and the stage. Brantmoor, at €1,562k a year, is above €250k, so the claim reader called for a full brief. Small letters get a cited standard reply, held until a person sends it.
“23 of 23 letters · small letters (impact under €100k) get a cited standard reply, held until a person sends it.”
The claim reader highlights each claim in the letter: resin +3.0 points, energy +2.5, labour +2.7, freight and packaging +1.3, US import tariffs +1.5. The should-cost agent tests each against the index the contract names. The letter's “resin prices rose 22% in March” was one spot week, reversed by May; clause 7.3 uses the monthly contract index, which is down 4.0% since January 2025. Annika agrees, changes or disagrees with each verdict herself.
The claim reader never decides a claim is false; it records what the letter says and where.
On the left, the agents report in plain sentences, each with its sources as chips that open the passage: the spend grew from €12.4M to €14.2M in two years, Kestrel & Vane is 14.8% of Brantmoor's €96.0M revenue, and Brantmoor's energy is hedged to the end of 2027. On the right, the brief is seven cards, from our position to the pushback. Above both, one next step.
Fact checker: “I re-derived all 46 numbers on my own. 45 agree. One does not: labour… I will not pick one. Which should the brief use?”
From the 2023 open-book cost structure and each index since January 2025, the base of the last increase: resin −1.64 points, labour +2.08, overhead +0.60, freight +0.54, additives +0.30, energy 0.00. The costs support 1.9%, or 2.5% at Brantmoor's own 13.4% labour figure. Annika chooses the national index and keeps 13.4% as the upper bound; her choice and reason go on the record.
“+1.88 points on a 100-point price. Brantmoor asks +11.0. At their own 13.4% labour figure the costs support 2.5%.”
The peer median is +1.8% across 14 anonymised moulded-polypropylene settlements in H1 2026, upper quartile +2.9%. Kestrel & Vane's own: Orsava +1.4% and Feldkamp 0.0%, both with symmetric resin clauses, and Linwood +3.2%, traded for an 18-month extension. Orsava's quote puts the HX-40 housing at €2.61 against Brantmoor's €2.84, marked internal only. Excluding a comparable takes Annika's reason code.
Kestrel & Vane is 14.8% of Brantmoor's revenue; Brantmoor is 41% of its polypropylene spend, and its plant runs at 71%. Orsava is approved on 22 of the 38 parts, and Kestrel & Vane owns 14 of the moulds under clause 11.2. Annika drags the slider to 40%: 14 parts, 11 weeks, €269k one-off, €545k a year avoided on that volume against the letter.
The leverage analyst never contacts an alternative supplier.
It argues only from the letter, Brantmoor's accounts and past meeting notes, and expects six pushbacks, most likely “Resin rose 22% in March” at 90%. Each has a prepared, cited answer and a tip. Annika answers in her own words or with the prepared answer, and each answer is scored for the evidence used and any concession given away.
“10 / 10 — You answered from evidence; you gave nothing away. Do not debate the spot market. Point to the clause you signed.”
The brief writer proposed opening at 0% with a symmetric resin clause, a 2.0% target from 1 Apr 2027 and a 2.5% walk-away, trading the 12-month extension that costs nothing. Annika sets them. Priya confirms the budget fit: the FY2027 plan assumed €426k, the target is €284k and the walk-away €355k. Victor approves with TG-01, within his €500k authority, and the brief is locked as v4.
Signature meaning: “I approve these negotiation targets within my authority.”
Annika tells the brief writer in her own words: 2.1% from 1 Apr 2027, a symmetric resin clause with a ±3% trigger, a tariff surcharge only on US deliveries, for the 12-month extension and a monthly rolling forecast. It is 0.1 point above target, inside the walk-away: €1,264k a year avoided against the letter, estimated. The brief writer drafts amendment 3 and the savings record, both held, and the saving goes to Priya to validate.
“Both are held: the amendment needs your signature and Brantmoor's, the saving needs finance.”
The same letter, seen by five of the people who carry it — what the work looked like, and what it looks like now.
Eight agents research, rebuild the price, check each other and play the supplier; the category manager sets the targets.
Reads every price-increase letter on arrival, splits it into the cost claims it makes, finds the reply window in the contract and pre-scores the request.
Pulls spend history, part prices, contract terms, the supplier's filed accounts and news, and flags anything older than 90 days.
Rebuilds the price from the open-book cost structure and the change in each cost index since the base date, including the drivers the supplier did not claim.
Re-derives every number in the brief from its source, without seeing the other agents' working, and escalates every disagreement to a person.
Finds comparable settlements and part prices and explains why each one is comparable.
Measures dependency both ways, the qualified alternatives, tooling, switching time and cost, and sets out the best alternative to a deal.
Plays the supplier: predicts its pushback from the letter, its accounts and past meetings, prepares a cited answer to each, and runs the rehearsal.
Assembles the seven cards, proposes opening, target and walk-away, ranks asks and gives, and drafts the reply, the contract amendment and the savings record.
Sets targets; finance checks, the approver signs. The agents propose; a named person decides.
Every screen has the same bar. The agent that knows answers in plain words, cites its sources, and acts — opening the card, the benchmark or the draft — but never decides for you.
What do Brantmoor's costs really support?
About 1.9%, not 11.0%. On Brantmoor's own 2023 cost structure, polypropylene is down 4.0% on the contract index, energy is flat and labour is up 10.4%. At their own 13.4% wage figure it is 2.5% at most.
What if labour is 13.4%?
At the national index (10.4%), labour adds 2.08 points and the costs support 1.9%. At Brantmoor's own 13.4% settlement it adds 2.68 and the result is 2.5%. That is why the walk-away is 2.5%: even on their figure, no more is supported.
Exclude Linwood
Linwood's +3.2% was traded for an 18-month extension, so it is a weaker comparable. I have opened the benchmarks with Linwood ready to exclude: pick a reason code and confirm. I will not drop it without you.
Draft the reply to Brantmoor
Drafted and held for you. It declines the 11.0% as notified, cites the resin index and clause 7.3, and asks for the open-book breakdown clause 7.5 requires. It gives away no target or benchmark.
13 screens from the working solution, on its sample data. Pick one to see it large.
The administrator sets who decides what and the rules the agents follow. Every change is versioned and on the record.
| Setting | Default | Choose from |
|---|---|---|
| Category manager approves alone up to | €100k a year of impact | Any amount; thresholds must rise in order; a change is a new version the CPO approves |
| Head of sourcing approves up to | €500k a year of impact | Any amount above the category manager's; the CPO approves above it |
| Answer every letter within the contract's reply window | On · 30 days when the contract is silent | On · off |
| Cited standard reply for small letters | On · impact under €100k a year, held until a person sends it | On · off |
| Full brief above €250k impact | On · seven cards, fact check and rehearsal | On · off |
| Walk-away at most the upper should-cost | On · a higher walk-away needs a written reason | On · off |
| Request an open-book breakdown where the contract requires it | On · drafted by the brief writer, sent by a person | On · off |
| Re-check sources older than 90 days | On · refreshed before the meeting | On · off |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample negotiation · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (a 2025 supplier study on tariffs, reported by Chain Store Age — 49% of suppliers planned tariff-driven increases; a 2025 GlobalData tariffs poll — 24% of companies not raising their own prices planned to negotiate with suppliers; a procurement training site's practitioner estimate of a full day to prepare a major supplier negotiation; CIPS guidance calls preparation the first and most important stage of a negotiation). People, companies, suppliers, contracts and index values named on this page are fictional — characters and sample data in the working solution.
AI procurement negotiation preparation for category managers. Eight agents read a supplier's price-increase letter, pull the spend, contract and the supplier's accounts, rebuild the price from the contract's cost indices, check every number twice, bring benchmarks and alternatives, and play the supplier in a rehearsal. The result is a seven-card brief with every number cited; the category manager sets the targets and the head of sourcing or CPO approves them.
The claim reader splits the letter into its cost claims and finds the reply window in the contract. The should-cost agent tests each claim against the index the contract names, from the base date of the last increase, and adds the drivers the supplier did not claim so the answer is fair. The brief writer drafts a cited reply, held until the category manager sends it.
It takes the supplier's open-book cost structure — in the sample, resin 41%, labour 20%, overhead 12%, energy 8% and so on — and applies each index's change since the last increase. On the sample letter that gives +1.88 points, so 1.9% supported against 11.0% asked. You can switch an assumption, such as the labour figure, and see the result move.
No. The agents never contact the supplier or an alternative supplier, never send a reply, never sign an amendment and never book a saving. Replies and amendments are drafted and held until a person sends or signs them.
Yes. The brief writer only proposes opening, target and walk-away. The category manager sets them, finance checks them against the budget, and the head of sourcing or the CPO approves them within the authority matrix with a reason code. The approved brief is locked, and a change needs a new approval.
The category manager tells the brief writer what was agreed in plain words. It measures the result against the locked target and walk-away, works out the cost avoided against the letter, and drafts the contract amendment and the savings record. A saving is booked only when finance validates it.
Yes. The approval thresholds, the reply rules — standard replies for small letters, a full brief above an impact, the walk-away limit, open-book requests, source freshness — and the index mapped to each cost driver are all settings, versioned and on the record.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your authority matrix, contract templates, index licences and a few recent price-increase letters — and it goes live once every quality gate has passed. We will walk you through it on one of your own letters first.
Hidden costs priced before award, and every score backed by the page it came from.
1-min demoSpend AnalyticsEvery dollar classified, and every saving pinned to the invoice lines behind it.
1-min demoRFP CreationHalf the bidder questions, because every gap past bidders asked about is closed first.
1-min demoWe'll run Supplier Negotiation Preparation on one of your own recent price-increase letters.