Today
Every morning someone builds the production report by hand from several systems and spreadsheets. It shows numbers but not why a line is behind plan. The reasons are then chased in the morning meeting.
An example
At Talwerk Antriebstechnik the 6:30 report shows line 1 behind plan. The agent gives two reasons: a two-hour material wait on order 7790 and a planned tool change. Both statements link to the booking.
Your data, wherever it lives
Today confirmations, scrap and downtime sit in the MES, orders and the plan in the ERP, the figures in spreadsheets someone fills in by hand every morning, and the reasons in shift notes and e-mails. demi reads, understands and links them per line, order and shift, with your rights. The AI writes the report with its reasons before the morning meeting instead of someone spending an hour gathering figures, and the managers decide where to step in.
Where the data lives today
- MES
- ERP
- Office files
- E-mails
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: the MES, ERP, spreadsheets and e-mail. demi replaces nothing and links confirmations, orders and plan, scrap and downtime bookings and the shift notes. Every figure is tied to line, order and shift.
- 2 · PermitPermission Layer
Each manager gets the view of their own area, plant management sees the whole plant, and figures someone has no rights to appear in no report. Every dispatch is recorded.
- 3 · ActIntelligence Layer
The agent computes plan versus actual straight from the bookings, not by estimate, and links deviations to the recorded causes. Its short commentary points every statement to its booking, and the report goes out at a fixed time.
How it runs
- 1
Overnight demi pulls the bookings from the last shifts.
- 2
The agent computes output, scrap and backlog against plan.
- 3
It links each larger deviation to a recorded cause.
- 4
Every manager gets their view by e-mail in the morning.
- 5
You ask follow-up questions in chat and get answers with sources.
What changes
No more building reports by hand
Deviations explained, not just shown
Everyone works from the same facts
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
+132 hfreed per year
Assumptions and calculation
demi writes the report on output, scrap and backlog with reasons; you discuss the deviations.
60 min → 30 min per case
How we calculate
22 daily reports × 60 min × 50% × 1.0 (manual) × 12 months = 7,920 min = 132 h
Value per year, before demi costs
Freed per year+132 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
+€8,580
gained per year
132 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 each manager get their own view?
Yes. The view follows the org chart: the supervisor sees her line, the production manager his hall, management the plant. You set content, time and recipients per role.
Can I ask follow-up questions?
Yes, in chat, for example: “Why was the assembly line slower yesterday?” The agent answers from the same bookings and links the source. It only sees what you are allowed to see.
What if a booking is wrong?
The report shows what is booked and never corrects anything silently. Implausible values, such as output above capacity, are flagged as a note. The fix happens in the source system, and the next report picks it up.
What if no reason was recorded for a deviation?
Then the agent says exactly that instead of guessing. You see where bookings are missing, which is often the first lever for better reports.
