Today
Spend is spread across plants, cost centres and free-text orders. Nobody sees the total per commodity. Bundling and negotiation opportunities stay hidden, and the management report is rebuilt in spreadsheets every time.
An example
Talwerk Antriebstechnik buys DKOL 12L hydraulic hose fittings in three plants from 5 suppliers at different prices, some of them ordered as free text. The agent groups the lines, shows the price range per supplier and calculates the annual volume for one joint tender.
Your data, wherever it lives
Purchase orders and invoices sit in each plant's ERP today, some still in legacy systems, alongside free-text orders, supplier lists in spreadsheets and price agreements in e-mails. demi reads, understands and links them into one shared commodity structure, with your rights. In hours instead of days the AI shows the same parts bought at different prices and the bundles worth tendering, and the head of purchasing decides what goes to tender.
Where the data lives today
- ERP
- Legacy systems
- 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: every ERP, spreadsheets and e-mail. demi replaces nothing and sorts purchase orders, invoices, items and suppliers from all plants into one shared commodity structure. Suppliers and items created twice are merged into one record each.
- 2 · PermitPermission Layer
Views follow the org chart: plant buyers see their plant, the head of purchasing sees everything. The head of purchasing decides which bundle goes to tender, and corrections to the classification are recorded with person and date.
- 3 · ActIntelligence Layer
The agent classifies free-text lines too, totals spend per commodity and supplier and finds identical parts bought at different prices. Totals are booked invoices, not estimates, and every bundling proposal shows the lines behind it.
How it runs
- 1
demi reads purchase orders and invoices from every plant and system.
- 2
Every line lands in a commodity, free text included.
- 3
Identical parts and duplicate suppliers are merged.
- 4
The agent shows bundling potential with volumes and price differences.
- 5
Purchasing reviews the proposals and plans the tender.
What changes
Transparent spend per commodity
Bundling opportunities found
A management report you can back up
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
+77 hfreed per year
Assumptions and calculation
demi classifies spend across plants and systems and shows where bundling pays; you negotiate.
16 h → 9.6 h per case
How we calculate
1 spend analyses × 16 h × 40% × 1.0 (manual) × 12 months = 77 h
Value per year, before demi costs
Freed per year+77 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
+€5,005
gained per year
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 it classify free-text orders?
Yes. The agent assigns free text to a commodity using description, supplier, cost centre and similar lines. Uncertain assignments are marked. Every correction by purchasing improves the next assignments.
Does it work across several ERPs and plants?
Yes. How the data gets to demi, we work out in a call. demi brings it into one shared structure. The systems stay as they are; nothing is replaced.
How reliable are the figures for management?
Totals are booked invoices, and every figure can be traced to the single line. Purchasing can review and correct classifications. You get a report you can back up line by line in the meeting with management.
How are savings documented?
Every bundling proposal records baseline prices, volumes and suppliers. After the award, demi compares the new purchase prices with that baseline. So it is clear where a saving comes from.
