Commercial decision engines

Analysis is not a decision.

A decision you can put your name on — and still defend in six months.

Everyone has AI now. Almost nobody has it where the money is decided.

It drafts the email, summarises the deck, cleans the data, writes the code. That is breadth, and breadth buys no advantage — your competitors bought the same tools in the same quarter at the same price. Depth is the lever, and depth means the handful of decisions each year that actually move the P&L.

01The price you set

and whether you can hold it when a competitor moves

02The supplier you award

and whether the rationale survives an audit

03The bid you price

and whether winning it was worth it

04The strategy you choose

enter or exit, build or buy, expand or hold — and how much to commit now

See it on your decision Run one free ↓ Two are live. About five minutes each. No sign-up.

Why not just ask an AI

A model can produce a recommendation. It cannot be accountable for one.

Because you have to defend it

Six months from now someone asks why you committed the money. You have the alternatives you considered, the numbers you used, the assumptions you made and who approved them — not a chat you cannot reproduce.

Because it has to hold up twice

Ask an LLM the same question two ways and the recommendation can move. Two people on your team, same facts, two answers, and the winner is whoever wrote the better prompt. Here the argument is about the assumptions — which is the argument worth having.

Because you need to know where to spend next week

You get the decision and what it turns on: which assumption would have to be wrong to change the answer, and by how much. That tells you what to go and find out, and what to stop researching.

Because it should still be there next quarter

The reasoning stays with your organisation, not with the analyst or the consultant who leaves. The next decision of the same shape starts from something.

Where the AI stops

The AI reads. It never decides.

The AI does a great deal of the work. It reads, researches, interrogates and argues. It is never the thing that sets a number.

01 · The AI

Reads your documents

Tenders, RFPs, vendor submissions, cost files. It pulls them into structured inputs instead of you retyping them.

02 · The AI

Finds the evidence

Live market and counterparty research, run inside the bench. Every finding arrives with a source you can open and check.

03 · You

Confirm every input

Nothing computes until each assumption is in front of you. A model that quietly supplied its own numbers would make the arithmetic reproducible and the answer not.

04 · The engine

Does the arithmetic

Same inputs, same answer, every time. Two people with the same facts get the same recommendation — not whoever wrote the better prompt. No AI touches a number here.

05 · The AI

Attacks the result

It red-teams the case, runs consistency checks against it, and will play your counterparty across a full negotiation before you meet the real one.

06 · Beyond it

Escalates, never guesses

Where a decision falls outside what the engine can resolve, it says so. The override is yours, recorded with its grounds. The objective it maximises is set by you.

Then it tells you how far each assumption can move before the recommendation changes.

That is the number the decision actually turns on — what to go and verify next week, and what to stop researching.

You hold the pen.

See the spine of this run in about five minutes — try one below ↓

Four engines

One method, applied to the four decisions above.

01 · The price you set

Pricing

What to charge, when to move and how to defend it. Three sub-decisions: pricing an enhancement you have just built, adjusting a price under cost pressure, and entering a new market.

You leave with a price, the revenue and volume consequences, and the competitor response it assumes. Two of the three are runnable now ↓

02 · The supplier you award

Procurement

Which supplier, at what total cost, under what conditions — with criteria scored before any supplier was evaluated.

You leave with an award decision your CPO can sign and your auditor can follow.

03 · The bid you price

Bid & deal

Whether to pursue, at what price, and when walking away beats winning — win probability quantified before the conversation starts.

You leave with a floor, a target and a concession plan.

04 · The strategy you choose

Strategic options

Entering or leaving a market, acquiring versus building versus partnering, restructuring a supply chain, expanding capacity, divesting a line, licensing rather than commercialising, funding a transformation in tranches — commitments made in stages, with evidence arriving between them.

You leave with the path to commit to, what each gate must establish, and the point at which that choice reverses.

Where the method comes from

Built from the pricing and procurement work, not from a product roadmap.

ClientsBuilt on pricing and procurement work for McKesson, Roche, Honeywell, Galderma, Kemin Industries, Great-West Retirement Services, Pioneer Hi-Bred (now Corteva), Cognos (now IBM), GEAC, Esselte, Duraflame, Deringer and Cuscal.
PracticePartner and principal in several consulting firms, including Harrison & Associates, SPMG Global, Oliver Wyman (Larson & Associates) and PricingCloud.
TeachingAuthor of a Harvard Business School case used in the first-year BGIE curriculum. Pricing workshops taught in Johannesburg, Paris, Singapore, Kuala Lumpur and San Francisco.
MethodRaiffa decision analysis, McFadden discrete-choice modelling, Nash bargaining — the three that sit under every engine here.
Reviewed bySenior partners and practitioners in global procurement and innovation consulting. Advisors →
DeliveryEngagements run directly, or jointly with partner firms who hold the client relationship and lead the work. Consulting practices also adopt the engines as a permanent in-house capability. How we work →
ContinuityEach engine is a single file that runs inside your own environment, with no external infrastructure. The method is codified in the engines and their documentation — nothing here depends on one person being available to operate it.

Alain Meloche

Founder and principal · Toronto

Twenty-five years of B2B pricing and procurement work, before any of it was software. Engagements are led personally and staffed with partner firms where the work calls for it.

Book a call