AI quoting saves the most time on the rebuild work: reading the RFQ or your site notes, finding similar past jobs, and drafting the line items. It doesn't know your margin, how booked you are, or which customer pays late. Let AI build the draft, and keep the price with you.
TL;DR
- AI can read the request, pull past jobs, and draft line items from your price list.
- You still set margin, rush and capacity pricing, and anything new or vague.
- Write your pricing rules down before you automate them.
Where does quoting time actually go?
A lot of it goes to work that isn't pricing. Someone opens the email, reads the drawing or the notes from the walk-through, digs up the last job that looked like this one, retypes the specs into a spreadsheet, and emails the customer about the three things they forgot to mention. The judgment call at the end, what to charge, can be the shortest part.
The rebuild adds up even at big companies with full estimating teams. In a UHY study of large North American automotive suppliers, "the typical labor investment per RFQ has grown from about 134 hours to 157 hours." The same study found that "internal missed deadlines remain the most common issue in RFQ preparation" (UHY). Your quotes are smaller, but the pattern may sound familiar: the quote is late because it's waiting on someone to find and retype information, not because the math is hard.
That's the part AI is good at.
What can AI quoting take off your plate?
Contractors are already using it here. In a Houzz survey of 601 U.S. construction and design firms, "estimates and proposals were the most common project-management application, used by 87% of respondents employing AI in that area" (Roofing Contractor).
A quoting system built around your business can:
- Read the request. Pull quantities, materials, finishes, dates, and delivery terms out of an RFQ email, a PDF, or your field notes, and drop them into your standard intake form.
- Flag what's missing. Spot that the request has no finish spec or no site access details, and draft the question back to the customer so it goes out today instead of after the quote stalls.
- Find similar past jobs. Pull the three closest jobs you've quoted, what you charged, and whether you won them.
- Draft the line items. Build the first pass from your own price list and labor rates, not from generic numbers.
- Format the quote. Put it in your template, with your terms and exclusions, ready for review.
What you get back is a draft that's mostly assembled and not yet priced. That's on purpose.
Where do you still have to price the job?
Anywhere the right number depends on something that isn't in the request. AI can draft these lines, but a person should set them.
| Part of the quote | What AI can do | What you decide |
|---|
| Materials and quantities | Pull them from the request and your price list | Check anything unusual |
| Labor | Estimate from similar past jobs | Adjust for this crew and this site |
| Margin | Apply your standard markup | Raise or cut it for this customer and this job |
| Timing and rush | Note the requested date | Price the rush based on how booked you are |
| Risk | Flag vague scope or missing details | Decide what to exclude, allow for, or walk away from |
| New kinds of work | Find the closest thing you've done | Price it yourself, because there's no history to learn from |
Two of these deserve extra care. Capacity pricing is a judgment only you can make: when you're booked out, you might quote high on purpose, and when you need the work, you might sharpen the pencil. And customer risk, like the client who pays at 90 days or keeps adding scope, often isn't written down anywhere the AI can read.
Why do AI quoting projects disappoint?
Because the tool gets bought before the rules exist. In a ServiceTitan survey of 1,020 commercial contractors, run by Thrive Analytics, "among contractors using AI, 59% report a positive impact, but just 15% report a significant positive impact with clear return on investment" (ACHR News). That gap between "it helps" and "it pays" is the one to close.
Three things can get in the way:
- The pricing logic lives in one person's head. The UHY study found "nearly 90% of respondents reported that staff turnover, retirements, or reassignments are impacting the RFQ process." If your best estimator retires, the AI can't learn what they never wrote down.
- The tool can't see your data. A quoting assistant that doesn't read your price list and past jobs is guessing, and you'll spend the saved time checking its guesses.
- Nobody measured the before. If you don't know your current turnaround time, you can't tell whether the new system cut it.
How do you get pricing rules out of your estimator's head?
Do this before anyone builds anything. It's worth doing even if you never automate a thing.
- Pull your last 20 quotes, won and lost.
- Sit with the person who priced them and have them talk through each one. Why this markup? Why that exclusion? Why did this one lose?
- Write down the rules you hear: markup by job type, minimum charges, rush fees, travel, the customers who get a different rate.
- List the exceptions separately. Those stay with a person.
- Write down your current turnaround time from request to quote sent.
That list becomes the spec for the system, and it protects you the day your estimator takes a vacation.
How should you roll out AI quoting?
Start with one job type that you quote often and price consistently. Run the AI draft side by side with your estimator for a few weeks, and compare the draft to what actually went out. Where they differ, update the rules or move that decision to the "you decide" column.
AI Marketing Labs builds this kind of system as Custom AI Automations, and quotes are one of the processes it covers. It's built on your accounts and your tools, and once it's handed off, it's yours with no ongoing fee to us. Contractors can pair it with the other places AI saves construction businesses the most time.
Once a quote turns into a yes, the job has to land on the calendar. AI appointment scheduling follows the same rule as quoting: automate what follows clear rules, and keep a person on the calls that don't.
What should you measure once it's running?
Pick three numbers and check them monthly:
- Turnaround time. Hours from request received to quote sent, compared with the number you wrote down before the build.
- How much the draft changes. If your estimator rewrites most of the line items, the rules or the price list need work before you trust the draft.
- Win rate by job type. Faster quotes are only worth it if they keep winning at prices you're happy with. If the win rate jumps on one job type, check if your price on that job type is now too low.
Also keep an eye on the questions the system sends back to customers. If the same missing detail comes up every week, add it to your request form so it stops holding up quotes.
Frequently asked questions
Can ChatGPT make a quotation?
It can format one from what you type in, but it doesn't know your costs, your rates, or your past jobs unless it's connected to them. Check every number before a quote from a general chatbot goes out.
Is AI quoting accurate?
It's as accurate as the price list and past jobs it draws from. Keep a person reviewing every quote, and track where the draft and the final price differ.
Does AI quoting work for custom jobs?
It helps with the intake and the line items even on custom work. The pricing on anything truly new should stay with you, because there's nothing in your history for it to learn from.
How long does it take to set up AI quoting?
It depends mostly on how well your pricing rules are written down. If they only exist in one person's head, start there, because that step takes longer than the build.
