Today
Every request lands in estimating the same way: the jobs you can win and the ones that will never fit your machines, your capacity or your price level. Estimators spend days on hopeless quotes while good requests wait. Whether to bid is often decided by gut feeling on Monday morning.
An example
On a Monday, Norvelta has eleven requests in its inbox. The agent recommends quoting four, among them a 2K housing mould, as three similar moulds kept their margin in actual job costing, asking back on two because tolerances are missing, and declining five: three moulds are too large for the biggest milling machine, and two come from a customer whose last six quotes were all lost on price. The sales manager goes through the recommendations in a quarter of an hour and passes the four to estimating.
Your data, wherever it lives
What you need for the decision is scattered today: requests and drawings in mailboxes and PDFs, actual job costs and order backlog in the ERP, win and loss reasons in the CRM, machine loading in the MES, earlier parts and tools in the PLM. demi reads, understands and links them, with your rights. The AI rates every request in minutes instead of after days in estimating, and whether you quote is your decision.
Where the data lives today
- E-mails
- PDFs
- ERP
- CRM
- MES
- PLM
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: e-mail, PDF, ERP, CRM, MES and PLM. demi replaces nothing and links each new request with similar earlier requests, quotes and orders, their actual job costs, the customer's win and loss reasons and the planned loading of your machines.
- 2 · PermitPermission Layer
Ratings, margins and win rates are visible only to sales management and estimating. An authorised person decides whether to quote, ask back or decline; a rejection goes to the customer only after their release, and the decision is recorded.
- 3 · ActIntelligence Layer
The agent reads request and drawing, checks feasibility against your machines and earlier parts and rates capacity, expected margin and chance of winning by your rules. Every recommendation names its reasons and the cases it rests on; figures come from the ERP, CRM and MES, not from the language model.
How it runs
- 1
The request arrives by e-mail, with drawing, quantities and requested date.
- 2
The agent checks feasibility against machines, materials, tolerances and earlier parts.
- 3
It calculates capacity in the requested period and the margin of similar jobs from actual job costs.
- 4
From the history with this customer it estimates the chance of winning and recommends: quote, ask back or decline.
- 5
Sales management decides; good requests go straight to estimating, rejections go out after release.
What changes
Estimators work on requests you can win
A fast answer to the customer, even when it is no
Decisions by fixed rules, with traceable reasons
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
+180 hfreed per year
Assumptions and calculation
demi rates every request for feasibility, capacity, margin and chance of winning; you decide.
60 min → 42 min per case
How we calculate
50 customer requests × 60 min × 30% × 1.0 (manual) × 12 months = 10,800 min = 180 h
Value per year, before demi costs
Freed per year+180 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
+€11,700
gained per year
180 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
Which criteria does demi use to rate a request?
Yours: typical ones are feasibility, capacity in the requested period, expected margin from similar jobs, chance of winning with this customer and strategic points such as a new industry. You weight the criteria yourself, and every recommendation shows how it came about.
How does demi know the chance of winning with a customer?
From your history: how many quotes to this or similar customers turned into orders, why lost quotes failed and how prices developed. If there are too few cases, the agent says so instead of inventing a number.
Does demi decline requests on its own?
No. demi recommends and, if you wish, drafts a friendly rejection with reasons. It goes out only once an authorised person releases it.
What happens to requests with missing information?
If tolerances, material, quantity or date are missing, the agent recommends asking back and drafts the questions point by point. As soon as the answer arrives, it rates the request again.
