Where AI helps in quoting, and where it doesn't
A quote has two halves. The numbers (materials, labor, markup) need your pricing, your suppliers and your experience. The words (what's included, what isn't, what you assumed, when you can start, what happens next) are where quotes slow down and where disputes start. AI is good at the second half and unreliable at the first. It doesn't know your costs, local labor rates or how long your crew takes. Keep the price yours.
The method: voice note to clean quote
- Record on site. Before you leave, record a voice memo walking through the job: measurements, materials, condition issues, access, what the customer asked for and what you'd recommend. Talk the way you'd explain it to your lead tech.
- Transcribe it. Most phones transcribe voice memos, and ChatGPT, Claude and Gemini accept audio or pasted transcripts in their apps.
- Run the scope prompt below to turn the transcript into a structured scope, exclusions and questions.
- Price it yourself in your estimating software or spreadsheet. If you use field-service software such as Jobber, Housecall Pro or ServiceTitan, paste the scope into its quote template so the job stays in one system.
- Check and send using the checklist below, with a short cover message the AI also drafts.
The scope prompt
You're helping a [trade] contractor write a customer estimate. Here's my site-visit transcript: [paste]. Write: 1) a scope of work as numbered line items in plain English a homeowner understands, 2) exclusions (what is NOT included), 3) assumptions I'm making (for example, existing subfloor is sound), 4) questions I should confirm with the customer before sending. Do not add prices, quantities or anything I didn't mention. Mark anything you're unsure of with [CHECK].
Write a short email to [customer first name] sending this estimate. Mention the one thing they cared most about ([detail]), when we could start ([date range]), and how to approve it. Friendly, plain, under 120 words. No pressure tactics.
From this scope, suggest how to split the work into three options (essential, recommended, premium) using only the items listed. Don't invent new work. I'll price each option myself.
For a longer list of prompts for bids, customer texts and paperwork, see ChatGPT for contractors: 25 prompts.
Checks before you send
- Every line item matches what you saw. AI fills gaps with things that sound standard but weren't in your notes.
- Exclusions are explicit. Permits, disposal, hidden damage, paint, electrical: whatever applies to your trade.
- No promises you haven't made. Start dates, warranties and "guaranteed" anything stay out unless you put them in.
- Prices and quantities came from you, not the AI.
- Licensing, permit and contract language follows your state's rules and your own contract, not AI wording. Have your attorney or trade association template set the legal terms once, then reuse them.
- No customer data you don't need. Leave out access codes, alarm details and payment information.
Make the follow-up part of the quote
Many quotes die from silence, not price. When you draft the quote, have AI draft the follow-up messages too: one for three days later, one for a week, and a final check-in. Save them as templates in your field software or CRM. Our lead and quote follow-up guide has a full sequence with stop rules.
What to set up once
- A standard exclusions list for your trade that AI can pick from.
- Your tone and sign-off, saved in the AI tool's custom instructions or a project, so every draft starts in your voice.
- A price book in your estimating or field software, so pricing is fast and consistent.
- A rule for the team: AI drafts words, a person prices and sends.
Quoting is one workflow. Your business's AI plan maps it alongside calls, scheduling and invoicing for your kind of business. It names the tools you already pay for that can do more, and gives a monthly cost and setup time for anything new.