Today
Whether a batch was heat treated correctly sits in three sources: the furnace chart from the control system, the lab's hardness test and the furnace's pyrometry records. Anyone who wants to bring them together for one batch searches CSV exports, folders and spreadsheets, so day to day usually only the hardness gets checked. A soak time that was too short, an expired thermocouple or an overdue uniformity survey is found by the auditor, or by the customer when parts break in service.
An example
At Rabenhammer Präzisionsteile, shafts W Ø 40 h6 in 42CrMo4 are first quenched and tempered to 28 to 34 HRC, then induction hardened to 58 HRC at the bearing seats. For batch H-4471 with 380 shafts the agent overlays the charts of chamber furnace 2 and tempering furnace 1 on the programme: during tempering the temperature sat 18 °C below setpoint for 35 minutes because a radiant tube had failed. The hardness test shows 36 HRC on 2 of 5 samples. The agent proposes holding the batch and tempering it again, and also reports that the uniformity survey for chamber furnace 2 is due in 9 days. The quality manager holds the batch, the heat treatment shop re-tempers it, and the retest comes in at 31 to 33 HRC.
Your data, wherever it lives
What happened in a heat treatment batch sits in three places today: furnace charts in the control system, as CSV from a USB stick on the network drive or as paper strips in a binder, hardness values in the lab, the CAQ or spreadsheets, calibration certificates and furnace tests as PDFs. demi reads, understands and links them per batch and per furnace, with your rights. The AI checks every batch as soon as it leaves the furnace rather than at the audit, and quality decides on release or hold.
Where the data lives today
- Machines & sensors
- Folders & drives
- Scans & paper
- PDFs
- QM system
- MES
- ERP
- Office files
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: furnace charts and process data from the control system, often as CSV exports on the network drive, batches and jobs in the MES and ERP, hardness tests from the lab and CAQ, calibration certificates, uniformity surveys and system accuracy tests as PDFs or spreadsheets. demi replaces nothing and links, per batch, furnace, programme, temperature curve, soak time and quench with the hardness of the samples and the test status of each furnace. For every batch you see whether process, result and furnace were in order.
- 2 · PermitPermission Layer
Furnace data and hardness values are visible to the heat treatment shop and quality; customers see only the released records for their parts. Only the authorised person in quality may release or hold a batch, and every decision is recorded.
- 3 · ActIntelligence Layer
The agent overlays each furnace chart on the specified programme: heat-up, temperature, soak time, atmosphere, quench and tempering, each with its tolerance. It checks hardness values against drawing and specification and tracks the due dates for uniformity surveys, system accuracy tests, calibration and thermocouples under CQI-9 or AMS 2750. Limits and due dates are computed from programme, standard and furnace list; the language model only explains the deviation.
How it runs
- 1
After each batch, furnace charts and process data come from the control system, as a file or a scan.
- 2
The agent overlays them on the specified programme and flags every deviation in temperature, soak time, atmosphere and quench.
- 3
It checks the samples' hardness values from the lab or CAQ against drawing and specification.
- 4
It tracks the due dates of every furnace and flags uniformity surveys, system accuracy tests and calibrations in good time.
- 5
Quality releases the batch, holds it or orders it to be re-treated.
What changes
Every batch with a complete record from furnace and lab
Deviations show up before shipment, not in the audit
Furnace tests flagged before they fall due
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
+192 hfreed per year
Assumptions and calculation
demi compares furnace chart and hardness values with the spec; you decide on release or hold.
15 min → 9 min per case
How we calculate
160 furnace batches × 15 min × 40% × 1.0 (manual) × 12 months = 11,520 min = 192 h
Value per year, before demi costs
Freed per year+192 h
Assumption: labour cost of €49.50 per hour in German manufacturing in 2025 (Destatis), plus workplace, IT and overheads.
= Value per year, before demi costs
+€12,480
gained per year
192 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
Can demi read furnace charts from paper or CSV files?
Yes. Besides process data from the control system, demi reads CSV exports from a USB stick, scanned recorder strip charts and PDF printouts from older furnaces and turns them into temperature and time per programme step. If a chart is illegible or incomplete, the agent says so and does not assess the batch.
Which rules for furnace tests does demi know?
The CQI-9 and AMS 2750 requirements for uniformity surveys, system accuracy tests, calibration and thermocouples, plus your own rules and those of your customers. Each furnace's class, instrumentation and intervals sit in a list from which the agent computes the due dates.
How does the hardness test get matched to the right batch?
Through the batch number from the MES or ERP that appears on furnace chart, sample tag and test record. If the number is missing or the sample count does not match, the agent flags the gap instead of releasing the batch.
Does demi release a batch on its own?
No. The agent prepares the record and a proposal: release, hold or rework. The authorised person in quality decides and signs, and only then does the inspection certificate go to the customer.
