Analytics Ascent
Contact us

You have quoted this part before.

Give it the part you are pricing today and it finds the closest things in your quote history, tells you what they went out at, and says plainly when there is nothing close enough to be worth trusting.

95.0%
of a part's true siblings reach the top ten
±36.5%
median error on the price anchor
1,000
held-out parts it was measured on

Everything below runs in this browser tab. The catalogue is downloaded once and the matching happens on your machine, so a file you drop in never reaches a server — not ours, not anyone's. There is nothing to sign up for.

Matching against 0 sample quotes.

Take it apart

A demo built on a catalogue nobody can inspect proves nothing. The generator, its seed, the cost model, the tuning and the benchmark are all here. Change the dimensions, break the grades, re-upload it, and watch the match bands and the guardrail respond.

The catalogue is synthetic

No open dataset of real machined parts with real prices exists, so the sample here is generated from a published cost model. It is good enough to prove the matching works and to be benchmarked honestly. It is not evidence about your prices — that is what uploading your own history is for.

Carbon figures are indicative

Material-only, cradle-to-gate, from published emission factors. Accuracy is ±30% or worse depending on region and recycled content, and machining energy is not counted. Useful for comparing one grade against another. Not a footprint, and not ISO 14067.

Where your data lives

In your browser, for as long as the tab is open. Nothing is uploaded, stored or logged, and closing the tab is the deletion step. If your quote history cannot leave the building at all, the same engine runs offline as a script under NDA.

Run it on your own quotes

Drop a file above and it works today, against your history, in your browser. What a pilot adds is the part that cannot be done in a tab: tuning the weights on your data, parsing your descriptions properly, and reading quotes out of the system they already live in. Your data is in SAP, Odoo, Exact or Excel — we connect the one you actually use, once we have seen it.

Book a 20-minute call

Questions people ask

What is a should-cost model?
A should-cost model estimates what a part ought to cost from its physical and process characteristics — material, size, tolerance, finish, quantity — rather than from a supplier's quoted price. It gives a buyer or estimator an independent reference point before negotiating.
How do you build a should-cost model from quote history?
Instead of modelling cost from first principles, match the new part against parts already quoted on attributes known at enquiry time, then anchor the price to what those comparable parts actually went out at, adjusted for quantity and for the time since they were quoted. The history is the model.
Is this a machining cost calculator?
Not in the usual sense. A machining cost calculator builds a price upwards from assumed cycle times, machine rates and material cost, and its answer is only as good as those assumptions. This works the other way round: it finds the parts you have already quoted that are most like the new one and anchors the price to what those actually went out at. If you have no quote history the bottom-up calculator is the only option; if you have five years of it, your own prices are better evidence than anyone else’s machine rates.
What makes two machined parts comparable?
Material family first, because a part in aluminium tells you nothing about the same shape in a nickel alloy. Then envelope dimensions sorted so orientation does not matter, part mass, tightest tolerance, feature count, number of setups, surface finish and batch quantity.
Why would a pricing tool refuse to give an answer?
Because a number produced from parts that are not actually similar is a guess wearing the costume of an estimate. When the nearest match is further away than anything the method was tested on, saying there is no reliable precedent is the more useful output.