Today
For many injection molds a paper setting sheet hangs in the tool cabinet, with crossings-out and notes such as “holding pressure +20 bar, sink marks gone”. The data set on the machine is often a different version, and when the mold runs on another machine, neither fits. So every start-up begins with trial and error, with scrap and with a call to the one setter who remembers.
An example
At Polivo Kunststoffverpackungen the 32-cavity mold for the VS-38 closure is due to restart after eleven weeks, this time on machine 14 instead of machine 9. The tool cabinet holds four setting sheets with handwritten corrections, the machine a data set from 2023. The agent shows that the June run was the best, with 0.4 % scrap: holding pressure 30 bar higher than on the sheet, cooling time 0.3 s shorter, noted by hand with “sink marks gone”. It converts metering stroke and pressures to the larger screw of machine 14, and the setter has good parts after 25 minutes; the last start-up took two hours.
Your data, wherever it lives
Good settings live today on paper sheets in the tool cabinet, as data sets on individual machines and USB sticks, as process data in the MES and as experience in the setters' heads, apart from scrap and inspection results in the QMS. demi reads, understands and links them per mold, machine and material, with your rights. The AI shows the last good setting in seconds before every start-up, and the setter decides at the machine what to apply.
Where the data lives today
- Scans & paper
- Machines & sensors
- MES
- QM system
- ERP
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: scanned setting sheets from tool cabinets and binders, data sets from the machine controls, process data and scrap from the MES, inspection results and defect records from the QMS and the order history from the ERP. demi replaces nothing and joins everything into one record per mold, machine and material. You see every run with its setting, result and note.
- 2 · PermitPermission Layer
Setting data are visible to setters, shift leads and process engineering; data of customer-owned molds stay with that customer's team. demi writes nothing into a machine control. A new standard data set is released by process engineering, for validated processes through your change procedure, and that is logged.
- 3 · ActIntelligence Layer
The agent transfers values and handwritten corrections into fields, unifies units and links every run with scrap and defect records. It shows the setting that last produced good parts and explains how it differs from other versions. Conversions to another machine follow fixed formulas from screw diameter and pressure ratio and are marked as a proposal.
How it runs
- 1
Setting sheets from tool cabinets and binders are scanned; machine data sets and MES process data are connected.
- 2
demi reads every sheet including handwritten corrections and assigns it to mold, machine, material and date.
- 3
For each run the agent links setting, scrap, defect records and inspection results.
- 4
Before the next start-up it shows the setting that last produced good parts and converts it to another machine if needed.
- 5
The setter applies the values at the machine, and process engineering releases the new version as a data set.
What changes
Start-ups begin with the last good setting instead of trial and error
Less start-up scrap and shorter changeovers
Setters' knowledge stays in the plant, across shifts and years
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
+120 hfreed per year
Assumptions and calculation
demi shows the last good setting for each mold and machine; the setter applies it at the machine.
40 min → 28 min per case
How we calculate
50 mold start-ups × 40 min × 30% × 1.0 (manual) × 12 months = 7,200 min = 120 h
Value per year, before demi costs
Freed per year+120 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
+$7,800
gained per year
120 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 handwritten setting sheets?
Yes, from a scan or a phone photo. demi recognizes the values in the form fields, crossings-out and margin notes and marks handwritten corrections as such. If a digit is not clearly legible, such as a 1 or a 7, the agent shows the cut-out and asks.
Where do the machine data sets come from?
From the machines' data set storage, from the MES or from the drive where your setters keep them. How the data reaches demi, we clarify in conversation. For older machines without an interface, a photo of the setting screens is enough.
Does demi set up the machine itself?
No. demi writes nothing into a machine control; the setter applies the values. Process engineering releases a new standard data set, and for validated processes, for example in medical technology or for automotive customers, every change goes through your change procedure.
How does demi know which setting was good?
From the result of the run: scrap from the MES or ERP, inspection results and defect records from the QMS, complaints about the lot. The agent shows what went well and badly in each run. What counts as good, such as a scrap limit or no complaint, you define.
