Today
Often only two or three senior planners know which machine really holds which tolerance. So new parts go to the machine they always ran on, not the one best suited for them. When those planners are away, even the feasibility check for an inquiry takes days.
An example
Rabenhammer Präzisionsteile receives an inquiry for a shaft W Ø 40 h6 in 42CrMo4, 620 mm long, hardened to HRC 58, bearing seats Ra 0.8. The agent rules out two turning centers because their turning length is too short and proposes the cylindrical grinder RS 2 in plant 2 for grinding: last year it held Ø 40 h6 with a Cpk of 1.67, the RS 1 only with 1.12. In week 49 the RS 2 still has 26 free hours, and production engineering takes the proposal.
Your data, wherever it lives
What your machines can do is spread across many places today: work envelopes and performance data in data sheets and spreadsheets, drawings in CAD and as PDFs, measured values in the CAQ, routings and free capacity in the ERP and MES. demi reads, understands and links them, with your rights. The AI checks a new part against every machine in every plant in minutes, and production engineering decides where it runs.
Where the data lives today
- CAD & drawings
- PDFs
- Office files
- ERP
- MES
- QM system
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
Data sources in this case: CAD, ERP, MES, CAQ and the file share. demi replaces nothing and links 3D models and drawings with machine master data, data sheets, clamping fixtures, inspection results and the capacity of every plant. For each machine you see what it can do on paper and what it achieves in practice.
- 2 · PermitPermission Layer
Customer drawings are visible only to the people allowed to work on the project. Machine and capacity data are visible to production engineering and production management. Production engineering decides which machine gets the part, and the routing goes into the ERP only after its approval.
- 3 · ActIntelligence Layer
The agent reads dimensions, tolerances, surfaces, material and weight from the model and drawing and compares them with the work envelope, power and achieved accuracy of every machine. It computes process capability from measured values of earlier parts and free hours from the schedule; the language model only explains why a machine fits or not.
How it runs
- 1
The inquiry or new part arrives with a 3D model and drawing.
- 2
The agent extracts dimensions, tolerances, surfaces, material and weight.
- 3
It compares them with the work envelope, power and fixtures of every machine in every plant.
- 4
Measured values of earlier parts show which machine really holds the tolerance; the schedule shows its free capacity.
- 5
Production engineering picks the machine, and the routing goes into the ERP after approval.
What changes
Every part on the machine that can really make it
Faster feasibility checks for inquiries
Machine know-how no longer in just a few heads
What does it bring you?
Check it with your own numbers
Every starting value is an assumption you can change. The calculation sits right below, step by step.
Your numbers
Result
+144 hfreed per year
Assumptions and calculation
demi matches drawing and tolerances against machine data and measured values; you choose the machine.
60 min → 42 min per case
How we calculate
40 feasibility checks × 60 min × 30% × 1.0 (manual) × 12 months = 8,640 min = 144 h
Value per year, before demi costs
Freed per year+144 h
Assumption: labor cost of $48.62 per hour in US manufacturing in June 2026 (BLS), plus workplace, IT and overhead.
= Value per year, before demi costs
+$9,360
gained per year
144 h ÷ 1,760 h per role = 0.1 full-time roles
An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.
Combine with more cases and get the packageQuestions about this use case
How does demi know what accuracy a machine really achieves?
From your measured values: inspection reports, first article reports and measurement data from the CAQ, per machine and feature. From these the agent computes the process capability of earlier parts. Where a feature has no measurements, it uses the data sheet value and marks it as such.
Does the machine selection work across several plants?
Yes. demi reads machine master data and schedules from every plant and shows all suitable machines in one list. Each plant keeps its rights: someone who sees only one plant only gets that plant's machines proposed.
Does this also work for a new format on a filling or packaging line?
Yes, on the same principle. The agent checks a new container or format against the performance data, format parts, cleaning rules and history of every line. It shows which line can run it and what must be converted or bought for it.
What if no machine can make the part?
Then the agent shows the closest machines and what each one lacks, such as turning length, spindle power, a fixture or a gauge. Whether you convert, outsource or invest, you decide with these figures.
