Automating quotes with AI, from the customer request to a reviewed draft
The request arrives by email, PDF or WhatsApp. AI extracts the items and quantities, applies the company price list and leaves a draft quote in the ERP. A person reviews it and sends it.
3 to 6 weeks
40 to 150 EUR / month
5 in the stack
The stack
Built with tools you already run
The problem
Quote requests wait for days until someone has time to do them.
A quote done by hand always follows the same steps: open the request, count quantities, look up prices in the price list or in old quotes, build the document in the invoicing software and send it. In construction, in a metalwork shop or in an auto repair shop, the person who knows how to do it is usually the same person who is on site, on the shop floor or at the counter.
Requests pile up, and the customer has asked other companies for the same thing. The part that needs knowledge of the business is small: setting the margin and reviewing anything out of the ordinary. The rest is collecting and copying data, which is what goes to process automation. When the quote has to come out of the company ERP, systems integration is what sets the project timeline.
The flow
How it works, step by step
The request arrives by email, form or WhatsApp
With attachments: PDF, photo, bill of quantities in Excel or a tender specification.
The request gets a number
It is logged with customer, date and channel before anything else happens.
AI extracts items and quantities
It reads the text and the attachments and returns a structured list, with anything unclear flagged.
Items are matched against the price list
List price or historical price, with the margin the company sets per customer type or job type.
The draft waits in the ERP for review
A person checks it, adjusts it and sends it. Nothing reaches the customer without that review.
Automatic follow-up
A reminder to the customer after a set number of days, which stops when they reply.
Walkthrough
The build in detail
A number for every request, from the start
It is what links the quote to the follow-up, to the order and, later, to the invoices for the job. Without a number, a request that came by email and another that came by WhatsApp for the same job become two quotes, and nobody can measure how many quotes the company wins.
Reading the request, attachments included
An AI model that reads documents extracts items, quantities and units from a PDF, a photo or an Excel sheet. Two checks prevent most errors: add up the extracted quantities and compare them with the total on the document, when there is one, and turn everything the request does not say (a missing measurement, the finish, the deadline) into a question for the customer instead of letting the AI assume.
The request describes build-ups, the price list has items
This is the pitfall that matters most in construction and building services. The specification asks for "plasterboard partition with two boards and insulation, 120 m²", and the price list has boards, studs, screws, mineral wool and labour. Before automating, you need a build-up table: each system, the items it is made of and the quantity of each per square metre. That is work for someone in the company, not for the AI, and it is done once.
With no price history, the estimate has nothing to stand on
The AI does not know what the company charges. If prices live in the head of whoever does the quotes, or are scattered across old quotes in PDF, the first step is to put them in one table, and the quotes and invoices from the last 12 to 24 months are usually enough. The AI helps extract them. When an item has no known price, the draft leaves it blank and flagged, and never makes up a figure.
Getting into the ERP
The quote should be created in the software the company already uses, so it has the right numbering, customer and items. Cloud software usually has an API. Software installed on the company computers, such as many versions of PHC, Primavera or Sage, may not, and then the integration goes through the database or an import file. That question sets the timeline, and it is answered in the first meeting.
Preliminary quote and final quote
For repeat work (the same repair, the same product in different sizes) the draft can be complete and the review is short. For bespoke work, the AI delivers a starting point with quantities and base prices, and the margin, the risks and the exclusions stay with whoever knows the job. Human review is a step in the flow, not an option.
What to expect
What this case delivers
Blueprint and video in preparation
We are preparing the .json file you can import into n8n, and the video walkthrough. Want this case running in your company already? Book a call.
Want this case in your company?
The production version includes integration with the systems you already run, monitoring, a fallback path and support. It is usually working within a week.
Book a free consultationNo commitment · Free of charge