AI is good at turning your rough notes into clear, friendly writing: quote descriptions, scopes of work, follow-up texts and review replies. It is not good at knowing your prices, your warranty terms or what your tech actually found. So the rule is simple: you supply the facts, the AI supplies the wording, and a person reads every draft before it goes to a customer.
Used that way, it can take a lot of the evening paperwork off your plate.
Why does wording matter on a quote?
Two quotes with the same price can get very different answers. The one that wins usually:
- Explains the problem in words the homeowner understands
- Says exactly what's included and what isn't
- Shows options, so the decision is "which one" instead of "yes or no"
- Sets expectations about timing, cleanup and warranty
- Sounds like a person who cares, not a parts list
Most owners know this. They just don't have the time to write it at 9 PM after a ten-hour day. That's the gap AI fills.
What should you give the AI, and what should you never let it invent?
Think of the AI as a fast writer who knows nothing about your business. It needs facts from you.
You provide:
- What the tech found, in plain notes
- The work you're proposing and the options
- Prices, from your pricebook
- Your warranty and terms
- Anything specific to this customer ("elderly, prefers phone calls")
The AI must never invent:
- Prices or discounts
- Warranty lengths or guarantees
- Licenses, certifications or years in business
- Findings the tech didn't report
- Deadlines or availability
If a draft includes something you didn't tell it, delete it. That's the most common way AI writing gets a business in trouble.
How do you write a quote description with AI?
Start with your tech's notes, however messy. For example:
"Water heater 40 gal gas, 14 yrs old, rust at base, small leak at drain valve, no expansion tank, code needs one. Options: replace like for like, upgrade to 50 gal, or tankless."
A useful prompt:
"Rewrite these technician notes as a short, friendly explanation for a homeowner. Explain what we found and why it matters in plain words. Don't add any prices, warranty terms or findings that aren't in the notes. Keep it under 120 words."
Then, for each option:
"Write a two-sentence description for each option below. Focus on what the homeowner gets, not technical specs. Don't mention prices."
You add the prices from your pricebook. The result reads like you sat down and wrote it carefully, and every fact in it is one you supplied.
Pair this with good, better and best options and you'll often see more customers choose an option than when they get a single take-it-or-leave-it number.
How do you write a scope of work for a bigger job?
Remodelers and contractors live and die by scope. A vague scope invites "I thought that was included" arguments later.
Give the AI your walk-through notes and ask for:
- A summary of the project in two or three sentences
- A numbered list of what's included
- A separate list of what's not included
- Assumptions (for example, "existing subfloor is sound; if not, a change order will be needed")
- What the homeowner is responsible for (clearing the room, choosing finishes by a date)
The "not included" and "assumptions" lists are where AI helps most, because they're tedious to write and easy to forget. Read them carefully. You're the one who knows whether a permit, disposal or patching is part of the job. Rules about what a written home-improvement contract must contain vary by state, so check yours.
What about follow-up messages?
Most quotes that don't close simply get forgotten. A polite follow-up a few days later brings many of them back. AI can help you write a small set of follow-ups once, which you then reuse.
Ask for three versions:
- Day 3: a short check-in with the link to the quote
- Day 7: a helpful note answering a common question, like financing or timing
- Day 14: a final, no-pressure message
Keep each one short. A text should fit on one screen. Then put them into your software's automatic follow-ups so they send themselves and stop once the customer decides.
How should you reply to reviews with AI?
Review replies are a perfect AI job: short, frequent and easy to put off.
For a positive review, give the AI the review and one detail from the job ("replaced their water heater on a Saturday"). Ask for a warm reply under 50 words that mentions the detail and doesn't sound like a template.
For a negative review, be more careful:
- Give it the review and the facts of what happened.
- Ask for a calm reply that acknowledges the experience, doesn't argue, doesn't share private details, and invites them to call you directly.
- Read it twice. Wait an hour if you're angry. Then post it.
Never let AI write fake reviews or reply in a way that reveals a customer's personal information. Google's review policies apply to you, not the AI.
How do you make AI writing sound like you?
The fastest fix is examples. Paste two or three messages you've written that you like and say "match this tone." Other tips:
- Tell it who's reading: "a busy homeowner" or "a property manager."
- Ask for plain words and short sentences.
- Ban the phrases you hate: "We're thrilled," "Don't hesitate to reach out."
- Ask for two versions and pick the better one.
A checklist for every AI draft
Before anything goes to a customer:
- Every price came from my pricebook.
- Every finding came from the tech's notes.
- Warranty and terms match what we actually offer.
- No promises about timing we can't keep.
- No private details of other customers.
- It sounds like us.
For using an assistant beyond writing, see how to use ChatGPT and Claude to run a service business. For the follow-ups that send themselves, see automations every service business should set up.
FAQ
Will customers know a quote was written with AI? Not if you give it real facts and edit the result. What customers notice is whether the quote is clear and specific. A generic, padded quote reads badly whoever wrote it.
Can AI work out my prices? It can help with the math, like markup and labor time, if you give it your numbers. It shouldn't choose your prices. Your costs, overhead and market are things you know and it doesn't.
Is it safe to paste customer details into an AI tool? Paste as little as you need. Most writing tasks don't need a customer's name, address or phone number at all. Check your AI tool's data settings.
Should my techs use AI in the field? It can help a tech turn quick notes into a clear summary. Keep pricing tied to your pricebook so techs aren't creating prices on the spot.
How ServFlo helps
In ServFlo, quotes are built from your pricebook, with good, better and best options customers approve online, so the prices are always yours. On Grow and Pro, Flo, the built-in AI assistant, can draft a quote from a plain description using pricebook items for you to review, and AI drafts replies to your Google reviews from the review and the customer's last job. Drafts stick to the facts they're given, and nothing posts or sends until you click. Quote follow-ups and invoice reminders can go out automatically and stop once the customer decides or pays. See quotes and invoices, reviews and marketing and Flo.
Photo by PuroClean of Fort Worth on Unsplash.