The short answer
Making quotes with AI means this here: the enquiry gets read, the data gets gathered, the standard wording and terms come from your own quotes, and a draft sits ready. A person sets the price and sends it. If you work with fixed packages and a configurator, you need no AI for this at all.
Example: from enquiry to a reviewed quote draft
Suppose a client emails asking for lunch on Friday for 35 people, including six vegetarian meals. The delivery address and preferred time are missing. The workflow extracts the quantities, date and dietary requirements and prepares questions for the two missing details.
The draft uses the current price list you designate. For a fixed price per person, code calculates the subtotal for 35 people. Delivery charges stay open until the address is known; an old quote is only a reference. The employee checks availability, dietary requirements, price and terms, fills in the missing details and sends after approval.
This illustrative example follows the approach in the BON Catering case. The same sequence appears in service requests for installation companies: complete the enquiry first, then assess the work and price.
Where the time actually goes
Not in typing. In gathering. What exactly did the client ask, what did we charge last year for comparable work, which terms belong here, are the right measurements and quantities included and who can check that. A two page quote costs ninety minutes because those ninety minutes consist of searching through mail, through old quotes and through a colleague memory.
That gathering is exactly what a workflow does well and what a chat window does not. A chat window waits until you supply everything.
Who does which step
| Step | What the workflow does | What the employee does |
|---|---|---|
| Enquiry arrives | Reads the mail or form and pulls out client, work, quantity and deadline | Nothing, unless something is missing |
| Gather data | Finds earlier quotes for comparable work plus the standard wording and terms | Adds what only lives inside someone head |
| Draft the quote | Puts the quote ready in your layout, with line items and missing data flagged | Reads it through and fills the gaps |
| Set the price | Shows what was charged before, with the margin alongside | Sets the price. Always |
| Send | Puts it ready in the folder or system | Approves and sends |
What AI may not do
Set the price. A model can calculate what comparable work cost last year, which is useful. It cannot know that this client is difficult, that the contractor is short on time, or that you want this job to get a foot in the door. Leave pricing to a model and you sell at an average. Your margin lives in that gap.
And the quote never leaves on its own. An employee approves and sends, because a quote is an offer that binds you. How that boundary gets recorded sits on the page about control and security.
What you have to supply
- Twenty to fifty quotes you sent over the past two years, rejected ones included.
- Your current price list or calculation basis, and who keeps it current.
- The terms that go out as standard, and which exceptions occur.
- Who approves a quote, and above which amount somebody else has to look at it.
When you need no AI for this
If you work with fixed packages and the price comes from a table. Build a configurator instead, or switch one on inside the software you already have; that is cheaper, faster and free of mistakes. AI only pays off once every enquiry differs and somebody genuinely has to work something out. If you write a handful of quotes per month, the answer is also no: building costs more than the work.
If the bottleneck sits in following up rather than drafting, this is the wrong solution too. A quote that goes cold because nobody calls back does no better with a faster draft.
What it returns and how to test that
Calculate it before you have anything built. Take the number of quotes per week, the time they take today and the hourly rate of whoever writes them. The calculator turns that into a net process value, including the time review keeps costing. If no convincing amount comes out, you stop there. The free process check returns fit, doubt or no-go in seven questions, and the page about service requests shows how this runs at an installation company.
Want to see it on your own work first? Send one enquiry as it came in, without names or customer details, and I work it out for free. This is the first step of your free AI plan.
Frequently asked questions
Can AI write a complete quote?
A draft yes, a sendable quote no. The wording, the line items and the terms come out well. Pricing, exceptions and the final yes stay with a person, because a quote is an offer that binds you.
What if the enquiry is incomplete?
Then the workflow flags what is missing and puts a draft ready with the gaps visible, rather than inventing something. The employee fills them in or asks the client. A gap beats a wrong number.
Does the model use prices from my earlier quotes?
Only if you supply them, and only as reference alongside the draft. The workflow shows what was charged before and for which work. What goes into this quote is decided by the employee.
What does a quoting workflow cost?
1,500 euro excluding VAT for this single process, built in ten working days, tested, with a walkthrough for your team and handover. A free AI plan first tells you whether your volumes justify it.
About the author
I am Joost, AI specialist and founder of Tamboer AI. I have worked with AI daily since 2022. I tested and developed applications at one of the largest banks in the Netherlands and now build AI workflows for smaller businesses, with a person who checks every outcome. More about me.
