The business OS · Buying & vendors

Three quotes, and
a reason on the file.

One inquiry goes out to the vendors who can actually supply it, their quotes come back through a portal they do not need an account for, and the comparison is arithmetic before it is an opinion — with the reason you picked whoever you picked written down beside the order.

One module, 3 products. Sold separately almost everywhere else as vendor management, an RFQ or e-sourcing tool and a purchase-order system.

What it covers

Inquiry to order, with the argument in between.

The part that usually happens over email and in a spreadsheet, where the quotes are compared once and the reasoning is lost.

An inquiry, not an email threadWhat somebody needs is raised as a record with its line items, so the request that started the spend is still attached to the order at the end of it.

Items that survive vendor spellingA master item list, with synonyms and per-vendor name mappings, so "M8 hex bolt" and "Bolt Hex M8 SS" are the same thing when you compare. Unused synonyms decay rather than accumulating for ever.

Vendors with a catalogueVendor records with contacts, categories and the items they actually supply, so an RFQ goes to the ones who can quote it rather than to everyone.

One RFQ, many vendorsA single request fans out to the vendors you pick, per item, and the responses come back into the same place instead of into an inbox.

A portal vendors do not sign up forQuotes are submitted through a tokenised link. No account, no password to reset, no chasing a supplier through a procurement suite they will never log into twice.

Arithmetic first, opinion secondThe comparison runs in three layers: the deterministic one (lowest per item, lowest total) is always computed, a weighted score sits on top of it, and the AI explanation comes last. The AI never decides on its own.

Your weights, not oursPrice, quality and delivery are weighted to whatever your business actually optimises for, and the three are forced to add up to a hundred so the scoring cannot be quietly rigged.

Split the award by lineThe cheapest vendor for the whole basket is rarely the cheapest for every line. Preview a split award, see the coverage each vendor gives you, and award line by line if that is what the numbers say.

It says how sure it isEvery comparison carries a confidence score and the factors behind it, low when the history is thin and higher as it builds — instead of presenting a guess and a recommendation with the same face.

The reason is on the recordWhy this vendor was chosen, whether it followed the recommendation or overrode it, and the override reason if a human disagreed. Six months later the question "why did we buy from them" has an answer.

Purchase ordersThe decision becomes a PO with its items, so the order, the quotes it beat and the inquiry that started it all hang together.

Performance measured, not rememberedPromised delivery against actual, delay days and rejection rate computed by the database rather than typed in, quality scores and response times — built from your own orders, so the vendor rating is evidence.

A published ranking methodHow the ranking works is a page in the product, not a secret. A vendor who asks why they lost can be told, and a buyer who is challenged can show the method.

Straight answers

The things people ask on the first call.

Do vendors need an account to quote?

No. A quote comes back through a tokenised link — the vendor opens it, sees only the items you asked them about, and submits. Nothing to sign up for, nothing to reset, which is the difference between three quotes and one.

Does the AI choose the vendor?

No, and it cannot. The deterministic comparison — lowest per item, lowest total — is always computed and always available. Weighted scoring sits above it and the AI explanation above that. The AI describes a decision; it never makes one on its own.

Can we set what "best" means?

Yes. Price, quality and delivery carry weights you set, and the three are constrained to add up to a hundred, so nobody can quietly tune the scoring to reach a chosen answer.

What if one vendor is cheaper on only half the list?

Award the half. A split award is previewed before you commit — you see what each vendor covers and what the split costs against a single award — and the purchase orders follow the allocation you chose.

How do we answer "why did we buy from them" later?

The decision is stored with the comparison it came from, whether it followed the recommendation or overrode it, and the override reason if a human disagreed. The audit question has a written answer rather than a memory.

Where do vendor ratings come from?

Your own orders. Promised delivery against actual, delay days and rejection rate are computed from the purchase orders rather than typed in, alongside quality scores and response times. It is measurement, not reputation.

Does it know our items by our names?

It learns them. A master item list carries synonyms and per-vendor name mappings, so the same part quoted under three different descriptions still lines up in one comparison. Synonyms that stop being used decay instead of accumulating.

Stop buying on the first quote.

Send the request to the vendors who can supply it, compare on your own weights, and keep the reason.

This is one module of the business OS, not a point tool. Procurement on its own does what is usually sold as three: vendor management, an RFQ or e-sourcing tool and a purchase-order system. The same seat opens Meetings, Mail, Chat, Drive, CRM, HR, Accounts, Projects, Procurement, Learning and Research — eleven modules covering about twenty-eight products' worth, on one login, one permission model and one ledger. You are not buying a procurement tool; you are putting the company on one system.

See pricing Book a walkthrough →