The shop you just bought can make parts your estimators never quote
How do we route every request for quote to the right plant after buying more machine shops?
Figures and rules on this page apply to
United States
Working somewhere else? The shape of the problem usually travels. The deadlines do not.
What this looks like
A customer sends a drawing for a large structural part. It reaches the plant the buyer has always used, which cannot hold the part. The estimator declines it or quotes it high. Two states away, a shop you bought last month has the right machine and has run a similar job. Nobody sees the match in time.
The numbers
Every figure here is someone else’s. Check them.
- 262,240people employed in US machine shops in May 2022
- 1,170cost estimators employed in those machine shops, about one for every 224 workers
- 333,200people employed in US machine shops, turned product and fastener making in July 2026 (preliminary)
- 366,700the same count in August 2018
Why it happens
It is not a people problem.
Every shop describes its own capacity in its own words. One lists machines by brand, another by bed size, a third keeps it in the owner's head. Past jobs sit in each shop's own system under its own part numbers. Nothing says which shop can make which part.
Why your current software has not fixed it
Because it was never built to.
The group plan is one system for every plant. That takes years, and the next shop is bought before it finishes. A shared capability list goes out of date the week it is written. So requests are routed by who knows whom, not by who can make the part.
Intelligence, plumbed in
A drawing with a title block, a material callout and a handwritten note is readable, and it can be matched to past jobs.
We build the layer that says what each plant can make, then read every request through it. Underneath it: a connector, one agreed meaning per field, and a test set scored on your own records.
How we make AI survive real data-
Connectors -
A semantic layer -
Evals you can check
However hard, whatever it is
Machine shops are one example. Bring the request that sat longest before anyone quoted it.
We wrote this about machining groups because they are buying shops quickly. The same layer works for any group whose plants each quote their own way.
-
01 We sit with you
Days where the work happens, not a workshop in a meeting room. We watch the job get done and write down the shortcuts nobody wrote down.
-
02 We read everything
Your data, your rules, your vendors and their documentation, and the published research on your sector. We report what is actually in there.
-
03 We break it to first principles
Not which tool fixes this. What is actually causing it, taken apart until we reach the piece that cannot be divided further.
-
04 Then we build
Weeks, not quarters. By this point we are not guessing what to build, and guessing is the thing that makes projects long.
What we build
Specific enough to argue with.
Four mechanisms, not four features. Each one is a thing that happens on its own, every day, whether or not anyone remembers to run it.
-
Requests for quote read on arrival, drawings and notes included, whichever plant receives them.
-
One agreed description of every plant's machines, materials, sizes and approvals, kept current from the plants' own records.
-
Each request matched to the plants that can make the part, with similar past jobs from any plant attached.
-
A draft quote from that job history for the estimator to check, and each new shop added in weeks.
How you would know it worked
Numbers in your own reporting, not ours.
- Share of requests quoted by a plant that can make the part first time.
- Days from request received to quote sent, per plant.
- Requests declined for lack of capacity that another plant in the group could have made.
Straight answers
Where a model is involved, it is scored against your own records first. Accuracy per source, not one flattering average.
The questions this raises
-
Our estimators know which shop does what.
They do, for the shops they grew up with. The layer puts that knowledge in one place, so a newly bought shop is visible on day one.
-
Some of our work is export controlled.
Then the layer runs where that data is allowed to live. Controlled drawings stay inside your own systems and approved hosting.
-
We plan to put every plant on one system. Why do this first?
The move needs the same map of machines and part numbers anyway. Building it first helps quoting now and shortens the move.
Where the figures come from
We did not make these up, and you should not take our word for them.
Also in manufacturing
Productivity fell in fourteen of the twenty manufacturing groups measured
Federal figures show labour productivity declining across most manufacturing industry groups. Machine data is abundant. The work between machines is not recorded at all.
ProblemEvery plant you add keeps its own production record
Contract device makers grow by buying plants. Each one arrives with its own travellers, lot numbering and record format, and the auditor still expects one traceable story.
ProblemEvery hotel project arrives as a different drawing set
Contract furniture makers win hotel and retail projects on bespoke work. Each brief arrives as someone else's drawings, schedules and specs, and quoting waits for a person to read them all.
ProblemEvery line you built is one of a kind, and its history is in service notes
Makers of packing and sorting lines build each one to order, then service it for years. The real record of each line sits in service reports and engineers' notes.
ProblemEach shop in the group quotes alone, from its own past jobs
Groups of contract manufacturers grow by buying local shops. Each shop still quotes from its own records, so the group never sees what it already knows.
ProblemOne heat network, two pipe ranges, and every bid costed by hand
Pipe makers now sell complete heat-network systems. Each bid still means reading someone else's network plans and costing every run by hand, often in two separate catalogues.
The meaning is not in any schema
Two systems agree on the field and disagree on what it means.
The rule lives in a document, or in what one site has always called something, or in a habit nobody wrote down. Joining the data is the easy half; agreeing what it means is the half that is actually the work.
- Behavioral healthOne in four facilities still run on paper, and there is a reason
- Accounting & consultingAn eight-hundred-billion-dollar quarter built on hours recalled from memory
- Senior livingNobody wants a sixth dashboard
- Skilled nursingEvery building reports. Nobody can see the portfolio.
- Supply chainThe distributor you just bought still takes orders its own way
Consent rules, site vocabulary, what an hour counts as. Wherever two systems agree on a field and not on what it means.Describe the disagreement.
Routing a quote across plants is a matching problem, not a new system.
We build the layer that says what each plant can make, then read every request through it.
See everything we build-
Software -
Hardware -
Ways of working -
Whole ventures
Is this happening to you? Tell us the size of it.
Twenty minutes. We will tell you honestly whether the numbers justify doing anything about it.