Case study · Trades
From 3 hours to 20 minutes per quote.
A heating and plumbing company in Bavaria with 15 employees. A website project grew into a company brain: the company's knowledge is stored centrally, and an agent uses it to answer questions and get work done in the ERP. Funded through the Bavarian Digitalbonus (Plus tier).
- Time per quote
3 h → 20 min
Time per quote
- Employees
15
Employees
- Digitalbonus Bayern
Plus
Digitalbonus Bayern
Starting point
Requests arrived through several channels: website, email, emergency calls. Knowledge about products, prices and procedures lived in people's heads and in folders. Every quote was put together by hand and took about three hours of office work.
The core
A company brain instead of a single bot.
The company's knowledge, meaning products, prices, procedures and documents, sits in its own knowledge base (RAG), separate from the AI model. The agent combines this knowledge with live data from Odoo: it answers questions and gets work done directly in the ERP. The knowledge belongs to the company and stays, even if the AI model is replaced.
What the agent handles
- Answering questions about products, prices and procedures
- Creating leads from requests directly in Odoo
- Providing information about invoices
- Preparing quotes in Odoo, reviewed and approved by a person
Quote preparation is one process of many. More are added step by step.
How the system fits together
From an incoming request to the reply to the customer, simplified.
Requests come in
- Website
- Emergency
AI agent
Understands the request and decides what needs to happen.
Company brain
The company's products, prices, procedures and documents.
Odoo ERP
Customers, leads, quotes and invoices.
Team reviews and approves
No quote leaves the company without human approval.
Customer
Receives a reply or a quote.
Technician app
Next step: the same knowledge directly on the job site.
What we built together
- 01
Website and lead management
An SEO-optimised website with about 80 pages. Contact, quote and emergency requests arrive in one structured place instead of scattered inboxes.
- 02
ERP with Odoo
One shared data foundation for customers, quotes and orders. Only then can an agent work reliably.
- 03
A company brain
A dedicated knowledge base (RAG) holding the company's knowledge, independent of the AI model. The model can be swapped, the knowledge stays with the company.
- 04
An agent that answers and acts
It draws on the company brain and on Odoo, answers questions and gets work done in the ERP, from creating leads to drafting quotes.
Result
A quote now takes about 20 minutes instead of about three hours. The project was funded through the Bavarian Digitalbonus at the Plus tier. Next comes a mobile solution for the technicians on site, drawing on the same company brain.
What we learned
The biggest lever was not the AI model but a clean process and a shared foundation of data and knowledge. Because the knowledge is stored separately from the model, it belongs to the company, not to the AI vendor.
We do not name the company for confidentiality reasons.
Another case studyTax firm: from incoming receipts to the annual statementA similar process in your company?
Describe briefly where people in your company search for knowledge or do work twice today. We will tell you honestly whether a company brain pays off and whether the project qualifies for funding.