Today
The board asks what a higher steel price, an expensive power contract or losing a major customer would mean for next year. Controlling builds a new spreadsheet, pulls data from three systems and spends a week on it, and by the time the answer arrives the question has changed. So decisions are often made on gut feeling, or only one scenario is calculated because there is no time for a second.
An example
Before the annual talks with its largest retail customer, the managing director of Tarvina asks: “What if rapeseed oil gets 25 % more expensive from March, power 30 %, and we only push through a 2 % price increase?” demi calculates across all 146 items: the margin of the deli salads falls by 2.6 percentage points, 17 items drop below the minimum contribution margin, and cash at the end of June is 380,000 euros below plan. Next to it sits the scenario with 5 %, in which only 3 items stay below, and she goes into the talks with both calculations.
Your data, wherever it lives
The figures for a scenario sit today in the ERP with bills of materials, recipes, purchase prices and sales, in the MES with run times and capacities, in the CRM with customer plans and in controlling's Excel sheets. demi reads, understands and links them, with your permissions. The AI runs a scenario in minutes instead of a week, with every assumption visible, and you decide what follows.
Where the data lives today
- ERP
- MES
- CRM
- Office files
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
This case needs data from the ERP, MES, CRM and controlling's spreadsheets. demi replaces nothing and links bills of materials and recipes with purchase prices, energy use per line, sales per customer and payment terms. Every raw material, every kilowatt hour and every customer is tied to the products it affects.
- 2 · PermitPermission Layer
Scenarios are run by management, controlling and the roles that already see margins and customer results today. A plant manager sees the effect on their plant, not the terms of individual customers. Scenarios change nothing in the ERP, and who ran which scenario with which assumptions is logged.
- 3 · ActIntelligence Layer
The agent turns your question into assumptions, such as a higher steel price from March for all suppliers, and shows them to you for confirmation. demi calculates the effect on margin, cash and capacity deterministically on real quantities and prices; the language model only explains where the differences come from. Several scenarios sit side by side, and every figure leads back to the data.
How it runs
- 1
You ask your question in plain language, for example: what if the steel price rises from March?
- 2
The agent turns it into assumptions with timing and scope and shows them to you for confirmation.
- 3
demi calculates the effect on margin, cash and capacity per product, customer and plant.
- 4
You compare scenarios side by side and change assumptions until the question is answered.
- 5
You decide on prices, purchasing or capacity, and you release whatever goes to the board or the bank.
What changes
Answers to scenario questions in minutes instead of a week
Several scenarios compared, not just one
Decisions with visible assumptions instead of gut feeling
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
+154 hfreed per year
Assumptions and calculation
demi runs scenarios on real ERP data with visible assumptions; you decide what follows.
8 h → 4.8 h per case
How we calculate
4 scenario questions × 8 h × 40% × 1.0 (manual) × 12 months = 154 h
Value per year, before demi costs
Freed per year+154 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
+$10,010
gained per year
154 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
Does the AI calculate the figures itself?
No. The figures come from a fixed calculation on your real quantities, recipes and prices. The language model only turns your question into assumptions and explains the result. Every figure can be traced back to the item, the booking or the customer order.
What data does a scenario need?
At least bills of materials or recipes, purchase prices, sales volumes and selling prices from the ERP. Run times from the MES add capacity, customer plans from the CRM add the effect of single customers, and payment terms add cash. demi shows whatever is missing instead of estimating it.
Can I change the assumptions myself?
Yes. Every assumption is listed with value, timing and scope, such as the new rapeseed oil price from March for all suppliers. You change it with a sentence or directly in the table, and demi recalculates in seconds. You save good scenarios and rerun them every month with fresh data.
Who sees the scenarios?
Whoever runs them and whoever you invite. Margins and customer results are visible only to roles that may already see them today. Management releases whatever goes to the board, the shareholders or the bank.
