243 files in. A checked bill of materials, routing and price out.
A metal fabricator prices every enquiry from the files its customers send. AI-Technolog reads the folder, checks it against the customer's own figures and prepares the routing and the price.
Built with AI-Technolog · Ready to deploy
Case study – Metal fabricator- ~40 hof a technologist's work saved on this one enquiry
- 0.0%difference from the customer's own weight total
- 8parts flagged for a person to check
- +1,174 kgof steel caught in a revision nobody had flagged
The problem
Every part analysed by hand, one by one.
Before there is a price, someone opens every model and drawing, one part at a time. They write out the material, dimensions and weight, work out the operations, and decide where each part will be made – on which machine in-house, or bought in.
The files come in the customer's structure, not yours: models, drawings, a parts list in Excel, revisions that arrive after the quote has started. On a 243-file enquiry that is around 40 hours of work, and the customer waits.
What we built
It breaks the job down into single parts and operations.
That is the core of AI-Technolog. Set up with the shop's own machines and rates, it does four things with every enquiry.
- 01
Reads the folder as it comes
Models, drawings, lists and revisions, in whatever structure the customer used. Nothing has to be tidied first.
- 02
Breaks the models down into parts and operations
From every model and drawing it takes each single part – material, dimensions, weight, quantity – and the operations each one needs: cutting, drilling, bending, welding, painting. The result is the bill of materials and the routing for the whole job.
- 03
Checks the customer's numbers
Part counts and weights recomputed from the models and compared with the customer's own – in total and per material.
- 04
Prices it, flags the unclear
Times from the shop's own machines, the price from its own rates – for every part and the whole job. Where the drawing and the model disagree, a person gets both numbers. Nothing is guessed.
The process plan · example
Where and how every part is made.
For each part, AI-Technolog proposes the route through the shop's own work centres, with a time. Bought-in parts are marked, and anything it cannot settle goes to a person with the reason. Below, four lines of it as an example.
| Part | Material | Route · work centre | Time |
|---|---|---|---|
| Main beam HEB 200 × 6 000 | S355 |
| 1.6 h |
| Base plate 300 × 300 × 20 | S355 |
| 0.4 h |
| Anchor bolt M24 × 500 | 8.8 |
Standard part, priced from the supplier list. | — |
| Bracing L 60 × 6 | S235 |
Drawing says L 60 × 6, the model says L 70 × 7. Left for a person, with both numbers. | ? |
A shortened example: four lines from the plan of a steel-structure job. The real plan had a line for every part of the 243 files.
The product · Ready to deploy
What you just read is AI-Technolog.
Drop in a customer's order folder. AI-Technolog returns the routing, times and price, checked against their figures, from your own machines and rates – only unclear lines go to a person.
How it works
From upload to a price the engineer signs off.
The engineer checks only the parts the system flagged.
- 01
Upload
The folder goes in as it arrived.
- 02
Check
Every figure the customer stated is recomputed and compared.
- 03
Route and price
Operations, times and the shop's rates give a price per part.
- 04
Review
The engineer confirms or corrects the flagged lines. Corrections are remembered.
The result
Every part counted, weight 0.0% off, and a revision caught.
On a steel structure of 243 files, every part was counted exactly and the weight landed 0.0% off the customer's own total, material by material. Eight parts went to a person, each with the reason.
Work that takes a technologist around 40 hours by hand came back ready for review. Their time went to the eight flagged parts and the sign-off, not to retyping drawings.
On another enquiry it caught a revision nobody had flagged: five beams changed to a heavier profile – 1,174 kg of steel that would otherwise have been priced at the old weight.
The same approach works for
- Quoting from customer drawings – sheet metal, machining, welding, steel structures
- Checking bills of materials and weights before an order is released
- Finding what changed between two revisions of the same job
Contact
A similar problem on your desk? Let's look at it together.
Thirty minutes with a founder: your process, your data, and whether a system like this one would pay for itself in your plant.
Or reach us directly
- Matyáš ProkopiusCo-founder
- Adam BulkaCo-founder