How to Use AI to Write Quotes and Estimates in a Small Service Business

AI can cut the hours behind a quote by reading the request, pulling your past jobs and price list, and drafting the scope and exclusions in the customer's format. It should never set prices or terms on its own. Prices come from your own price book, and a person approves every quote before it goes out.

Ask a contractor or a small shop owner where their evenings go and quoting usually comes up fast. The request arrives during the day, the owner or estimator is on a job, and the quote gets written at the kitchen table after dinner. Meanwhile the customer is collecting other quotes.

That's why quoting is one of the first jobs owners try to hand to AI. It's also one of the easiest places to get burned if it's set up badly. Here's what the 2026 data shows, which parts of a quote AI should do, and which parts it should never touch.

Can AI write quotes and estimates for a small business?

Yes, with a clear split of jobs. AI is good at the reading, finding and drafting that make up most of the time behind a quote. It is not a safe source of prices, discounts or terms. Those need to come from your own pricing rules, and a person should approve the final quote.

Mainspring, a UK company that sells AI quoting software to engineering firms, makes the case in a November 2026 post, Why quoting is one of the best jobs to give AI. It lists five traits that make quoting a good fit: it's reading-heavy, the answers already exist in past quotes and price files, it runs on rules, a person signs off, and the result is measurable because you either win the job or you don't. That's a vendor's argument, so read it as one, but the reasoning holds up for most service businesses I work with. (This Mainspring is a different company from the Mainspring that offers AI consulting in Texas.)

How many service businesses already use AI for quoting?

A lot of them, according to three 2026 surveys from field-service software companies:

• Housecall Pro. In its AI Advantage report, a survey of 248 home service pros in May and June 2026, 48% said they actively use AI. Among the 198 respondents who use or have tried AI, 51% use it for estimates, quoting and pricing, just behind customer communication and follow-up at 52%.
• Jobber. Its 2026 Home Service Trends Report, a survey of 1,050 US home service business owners in December 2025, found 52% use AI, and 54% of those adopters use it for quoting. Owners also named quoting as one of their biggest time wasters (37%), behind jobsite management and customer communication.
• JobTread. In its AI survey of contractors who connected AI to their JobTread account, published October 14, 2026, estimating was the most common use at 50%.

Read these with care. Each company sells software with AI quoting features, each surveyed its own audience, and the JobTread sample is customers who had already connected AI to their business software. They show that quoting is where a lot of trades businesses start with AI. They don't show how well it's working for them.

Which parts of a quote should AI do, and which must stay with a person?

Mainspring's second post, What does a quote really cost?, breaks the work into six stages: reading and qualifying the request, finding past quotes and prices, drafting the scope and answers, building the price, writing it into the customer's format, and handling clarifications. Most of that is finding and writing, which is where AI helps. The judgment calls should stay with you.

The good, better, best row is worth a look. Jobber reports that only 16% of the pros it surveyed offer tiered options, and that businesses offering optional line items see upsell rates between 25% and 50% on its platform. Drafting three options from your existing line items is exactly the kind of tedious work AI does well.

Why shouldn't AI make up prices?

Because a chat assistant will produce a confident, reasonable-looking number whether or not it has any basis. That's fine for a first draft of the wording. It's a real problem for a figure you're promising a customer.

Mainspring's position is blunt: prices "should never be generated by an AI model." They should be calculated from the business's own pricing logic and current supplier data, with the AI assembling the quote around them. It also says that when the AI's sources run out, the right behavior is to flag the gap and ask, not fill it with something plausible.

In practice, that means three rules:

1. Prices come from a price book. Your labor rates, material costs, markups and minimums live in your field software, a spreadsheet or a pricing tool. The AI pulls from that list and shows which line each number came from.
2. Nothing new gets invented. If a line item isn't in the price book, the draft marks it for a person to price.
3. Terms are fixed text. Payment terms, warranty language and validity dates come from approved wording, not from the AI's best guess.

Mainspring adds one more point I agree with: writing down how you quote often makes your quotes more consistent before any technology is involved.

Isn't pasting jobs into ChatGPT good enough?

It's a reasonable way to try the idea, and a poor way to run it. When each person pastes job details into their own chat account, you get different prompts, different prices and no record of what was sent. Customer names, addresses and job details end up in personal accounts the business doesn't control. Mainspring makes the same point about personal chat windows adding risk.

A shared setup fixes that. The AI built into the field-service software you already pay for, or a business AI account with your price book and standard wording, can work, as long as the whole team uses the same setup instead of personal chat accounts. If customer data is going into any AI tool, check the plan and settings first. I cover that in Is It Safe to Put Customer Data in ChatGPT? If you want the AI to read your job history and price list directly, How to Connect ChatGPT or Claude to Your Business Data explains the options in plain English.

How do you set up AI quoting in a small business?

Start with the tools you already have and one type of quote. Housecall Pro's own advice in its report is to save custom automations for later, because the built-in AI in the platform you run your shop on is often the foundation. Here's the order I'd follow:

1. Pick one quote type. Choose a common, repeatable job, like a water heater replacement, a standard service agreement or a repeat order, not your most complex bid.
2. Write down how you quote it. List your line items, labor rates, markups, minimums, standard exclusions and who can approve what. This is your price book and your rules.
3. Gather five to ten past quotes. Include a few you won and a few you lost. These show the AI your wording and level of detail.
4. Use the AI you already pay for first. Check what your field-service or estimating software can do before buying anything new.
5. Run it draft-and-approve. The AI drafts; a person checks every line and sends. Nothing goes to a customer automatically. I explain this guardrail in more detail in Your First AI Agent: A Small Business Guide.
6. Review weekly for the first month. Look at what the person changed on each draft. Repeated fixes tell you what to add to the rules or the price book.

Once quotes go out faster, make sure the follow-up keeps up. A fast quote that nobody follows up on still loses to the competitor who calls back. Speed to Lead covers the first-response side of the same problem.

How do you know AI quoting is working?

Measure before you change anything, then compare after 30 and 90 days:

• Hours per quote. Mainspring's method is simple: add up the hours each person spends on a quote and multiply by a fully loaded hourly rate (pay plus taxes, benefits and a share of overhead). Its examples are illustrative and in British pounds, so use your own hours and rates.
• Turnaround time. Days or hours from request to quote sent.
• Quotes sent per week. Are you bidding on work you used to turn down?
• Win rate. Jobber found 69% of the pros it surveyed report winning more than half their quotes. Track your own rate, not an industry average.
• Corrections. How many lines the person changed per draft, and whether any wrong price or term reached a customer. The goal is zero.

For a broader way to prove an AI project paid off, see How to Measure AI ROI in a Small Business.

How Caprock Agility helps

We don't sell quoting software. We help you figure out where your quote time actually goes and build a quoting workflow your team will use, on tools and accounts your business controls.

Our AI Readiness Audit ($2,500, 3–5 business days) maps how requests turn into quotes and jobs in your business, so you know whether quoting is the right workflow to fix first. If it is, EDGE Starter (starting at $8,500, 4–6 weeks) builds that one workflow end to end, including your price book and rules, the draft-and-approve process, team training and a Day 30 outcomes review.

You can compare every option on our services and pricing page.