How do you create quotes with AI?

Can AI create a quote for you?

Yes, AI can create a draft quote for you. It reads a customer's enquiry, puts the details in the right place and writes a first version of the text. Then a person looks at it. If everything is correct, the quote goes out.

That check matters. A quote is an offer. If the customer accepts the quote, for example by signing it and sending it back, it becomes a binding agreement. A mistake in the price or the schedule is then still your mistake, even if software made the draft.

AI here means generative AI: software that writes text based on an instruction. Such a system works with a language model. In this article you will read what AI does well with a quote, what a person still does, and how the whole chain from enquiry to invoice can work.

What does AI do well with a quote?

AI is good at work involving text. With a quote, it is mainly about these tasks:

• Reading an enquiry. A customer writes in an email or web form what they want. AI picks out the name, the address, the request and the preferred date.
• Spotting what is missing. Is a measurement, a photo or an address missing? Then AI prepares a question you can send to the customer.
• Writing text. A short introduction and a description of the work, in the tone of your business.
• Choosing standard wording. The right paragraph on delivery, payment or terms, suited to the type of job.
• Summarising a long attachment. For example a specification or a ten-page list of requirements.

This is work that comes up often and goes in a similar way each time. That is where AI fits best.

What does a person still do?

A person decides on everything that involves money or commitments. In practice, that means:

• Estimating. How difficult is the job? Do you need to go and look first? A tradesperson knows that better than software.
• Checking the price. Is the quantity right? Was a discount agreed? Are there extra costs, such as a call-out charge or disposal?
• Assessing exceptions. A customer with an unusual request, or a job you would rather not do.
• Sending. The quote only goes out once someone has approved it.

AI sometimes makes mistakes that look convincing. It can add a detail that was not in the enquiry. This is called a hallucination. Makers of AI software, such as Microsoft, also write that answers from generative AI are not free of errors. You have to assess them yourself before you send them to others.

Where do the prices in an AI quote come from?

The prices come from your own price list or calculation, not from the language model. That difference is important. A language model is good with language, but it is not a calculator and it does not know your prices.

That is why a good set-up works in two parts:

1. AI reads the enquiry and determines which items belong in it. For example: replace gutter, twelve metres.
2. Ordinary software looks up those items in your price list and calculates the amount using fixed rules.

Is an item not on the list? Then the system does not make up a price. It sets the line aside for a member of staff. That way you know where every amount comes from. And if you change a price in your list, the new price appears in the next quote.

What does the chain from enquiry to invoice look like?

A quote does not stand on its own. It is preceded by an enquiry, and followed by planning and an invoice. If you want to automate quotes, look at that whole chain. In many businesses, someone retypes the same data at every step. This is how the chain can look when the systems talk to each other:

1. Enquiry. A customer fills in the form on your website or sends an email. The details go into your system automatically, for example your CRM.
2. Draft quote. AI creates a draft with text, items and the prices from your list, in your own template.
3. Check. A member of staff adjusts what is needed and sends the quote.
4. Acceptance. If the customer says yes, the work goes into the planning, with the address and description from the quote.
5. Invoice. When the work is done, a draft invoice is ready in your accounting software, with the lines from the quote.

The steps are connected through API integrations. The whole is also called workflow automation. AI is only one part of it: reading and writing text. The rest is fixed rules.

You don't have to build the whole chain in one go. Often you start with one step, for example the enquiry that goes into your system automatically. Do you recognise that you keep retyping data? Then also read You retype the same data.

How does an AI quote work per sector?

Below are three made-up examples. They show how it can work. They are not my clients, and they deliberately contain no figures.

What do you need to create quotes with AI?

Above all, you need order in what you already have. This is the list:

• An up-to-date price list. In one file or system, with fixed names for items. Is your price list currently mostly in someone's head? Then this is the first step.
• Standard wording. Your introduction, your terms, your payment arrangements and how long the quote is valid.
• A few old quotes. As an example of how your quotes look and sound.
• Access to your systems. To connect them, access is needed to your quoting, planning and accounting software. This usually goes through an API.
• Agreements about checking. Who reviews the draft, and who sends it?
• Agreements about data. An enquiry contains personal data, such as name, address and phone number. Then the GDPR applies. If an AI service processes that data for you, you sign a data processing agreement with that party.

Ondernemersplein (the Dutch government's information portal for businesses) lists the standard elements of a quote. Think of your company details, the customer's details, the date, a quote number, a description of the work, the price, how long the quote is valid and a reference to your general terms and conditions. Software can fill in the standard elements. You check the description and the price yourself.

Can your current software already do this?

That depends on your package. Many quoting, CRM and accounting packages have templates, and sometimes AI features too. Look at those first. If your package does what you need, you don't need to have anything built.

Custom work mainly fits in these cases:

• Your data is in several systems that don't talk to each other.
• You have your own calculation rules that a standard package doesn't know.
• The step from quote to planning and invoice is currently done by hand.

Is it mainly about two systems that need to share data? Then read what an API integration costs.

When is AI for quotes not worthwhile?

AI for quotes is not worthwhile if you make few quotes, or if every quote is completely different. A few situations:

• You make a few quotes a month. Then a good template is often enough.
• Every job needs a visit and its own calculation. AI can then at most prepare the text.
• You don't have a fixed price list. Then start with the price list. The rest comes later.
• You write large tenders with many requirements. Then it remains mostly work for people, with AI at most helping with summarising.

What does it cost to have quotes created with AI?

That depends on how you create quotes now. A fixed price list and a few types of work is less work than lots of custom jobs and exceptions. The number of systems involved also counts, such as your planning and your accounting. With me, you start with a business analysis. I look at how a quote is created now, from enquiry to invoice, and where the work lies. After that, you get a proposal with a fixed price.

The proposal states what I will build, for example draft quotes from enquiries, connected to your systems. Sometimes an integration between two systems is enough, for example from quote to accounting. It also states what is added each month for hosting and maintenance. That way you know in advance where you stand. Also read how your price is determined.

How do I help you with quotes and AI?

I build the chain from enquiry to invoice in the systems you already have. First I map how things work now. Then I make a proposal, and after that I build it myself. A person keeps checking the draft, and I explain to someone in your team how it works.

I don't receive commission from suppliers. That is why I first look at what your current software can already do. Read more about integrations and automation, or schedule a call. Do you mainly recognise that quotes take too much time? Then read Creating quotes takes too much time.

Frequently asked questions

Sources: Ondernemersplein: Step-by-step plan for creating a quote (in Dutch) (when a quote is binding and what it contains), Microsoft Learn: Data, privacy and security for Microsoft Copilot (in Dutch) (assessing answers from generative AI yourself) and Dutch Data Protection Authority (Autoriteit Persoonsgegevens): Data processing agreement (in Dutch). Consulted on 7 October 2026.