ChatGPT in a contractor's office: 12 jobs it does well, with prompts
Twelve office jobs where ChatGPT saves a contractor real time, each with a prompt you can copy and what to check before you send. Plus what not to paste in.
On this page
How many contractors actually use this?What should you never paste in?What are the 12 jobs?What does a before and after look like?What to do about it this weekSay you run a roofing and gutter company with a crew of six. This is a made-up example. It’s 9 p.m., the day’s jobs are done, and you still have four estimates to follow up on, three reviews to answer and a hiring ad you’ve put off for a month. None of it is hard. All of it is writing, and writing is what you’re slowest at after a day on a ladder.
That’s the kind of work ChatGPT does well: a first draft you edit. This post is twelve office jobs where it earns its keep, each with a prompt you can copy and a check to run before anything goes out. The prompts are plain text, so they work in Claude or Gemini too.
How many contractors actually use this?
Depends on who you ask and how they ask. The U.S. Chamber of Commerce reports that 66% of U.S. small businesses now use AI, and 41% use a gen AI chatbot. That comes from an online survey of 3,732 businesses with fewer than 250 employees, run with Teneo Research in July and August 2026.
The U.S. Census Bureau’s Business Trends and Outlook Survey gives a much lower number: 19.8% of all firms used AI in the period ending May 3, 2026. Census says less than 20% of firms with four or fewer employees reported using AI, and use among firms with fewer than 20 employees didn’t change significantly between December 2025 and May 2026.
The two surveys differ. Census asks about AI use “in any business function” over the past two weeks. The Chamber’s “small” goes up to 249 employees. The gap likely reflects those differences. Neither survey is a contractor survey. What both tell you is that plenty of businesses your size haven’t made this part of the office yet.
What should you never paste in?
On a personal ChatGPT account, OpenAI can use your conversations to improve its models unless you turn that off. To turn it off, open Settings, then Data Controls, and switch off “Improve the model for everyone.” Temporary Chats aren’t used for training and are deleted within 30 days. Even with training off, keep private details out of what you type. Leave out:
- Customer phone numbers, emails and street addresses (use “the customer” and “the Johnson job” style stand-ins)
- License numbers, bank details, payment links and passwords
- Contracts and bids that carry a customer’s or competitor’s private numbers
What are the 12 jobs?
For each one, copy the prompt, fill in the brackets, and read the output before you use it. The checks matter more than the prompts.
1. Estimate follow-up text
Write a short, friendly text to a customer who got an estimate from us [number of days] days ago for [job]. Ask if they have questions, don’t pressure, mention we can hold their spot on the schedule until [date]. Under 50 words.
Check: the date and the offer are true, and it sounds like you.
2. Reply to a good review
Write a two-sentence reply from the owner to this review: [paste the review text]. Thank them by first name, mention one specific thing they said, no marketing language.
Check: nothing is promised that you won’t deliver.
3. Reply to a bad review
A customer left this review: [paste]. Draft a calm public reply that thanks them, doesn’t argue or blame, says we want to make it right, and invites them to call [office phone]. Under 80 words.
Check: the facts. If the draft admits fault you haven’t agreed to, cut it. Our review rule post covers what you can and can’t do with reviews.
4. Scope of work from rough notes
Turn these job notes into a clear scope of work a homeowner would understand, as a bulleted list: [paste your notes, without the customer’s name or address].
Check: every line against what you’ll actually do. Add exclusions yourself.
5. Customer update email
Write a short email updating a customer that [what happened] and the new expected finish date is [date]. Be straightforward about the delay and say what happens next.
Check: dates and the reason for the delay.
6. Thank-you note after a job
Write a brief thank-you email after we finished [job type]. Include care tips for [the work] and ask if they’d leave a Google review, with a place to add the link.
Check: the care tips match your product. Make sure the review request follows the rules.
7. FAQ answers for a service page
I’m a [trade] in [city]. Write plain-language answers to these five questions customers ask: [list]. Two to three sentences each. Don’t invent warranty terms or numbers.
Check: every fact, number and warranty claim. Delete anything you can’t back up.
8. Google Business Profile post
Write a 60-word Google Business Profile post about [recent job or seasonal reminder]. No hashtags, no exaggeration, end with a reason to call.
Check: the claims, and that it’s a real job you did.
9. Hiring ad
Write a job post for a [role] at a [trade] company in [city]. Include duties, what we expect on day one, and what we offer: [list your actual benefits]. Plain language.
Check: don’t let it invent pay or benefits. Add your numbers yourself.
10. Summary of a long email thread
Summarize this thread in five bullets, then list what I need to do and by when: [paste thread with names removed].
Check: read the original for anything the summary dropped, especially dates and amounts.
11. Crew toolbox talk outline
Draft a five-minute toolbox talk outline on [topic] for a residential [trade] crew: three key points and two questions to ask the crew.
Check: safety content against your own program and the rules that apply to your trade. This is a starting outline, not a safety document.
12. Polite past-due reminder
Write a polite, firm reminder to a customer whose invoice for [job] is [number] days overdue. Reference the invoice number, ask for payment by [date], offer to answer questions, no threats.
Check: the invoice details and your contract terms. Have your attorney review anything that mentions fees or legal steps.
What does a before and after look like?
Here’s one from the made-up roofing company. The first version is what you’d type at 9 p.m. The second is what the prompt in job 1 returns after a quick edit.
Before: “Hey, just checking on the quote.”
After: “Hi Maria, it’s Dan at [company]. Just checking whether you had any questions about the gutter estimate. No rush. We can hold a spot on the schedule for you until Friday if that helps.”
The second one is longer and warmer, and it gives the customer a reason to answer. It still needs your edit: check the name, the offer and the date before it goes out.
What to do about it this week
- Pick the two jobs from the list that you do most often. Estimate follow-up and review replies are the easy starts.
- In ChatGPT, open Settings, then Data Controls, and decide whether to switch off “Improve the model for everyone.”
- Run each prompt once with a real case, without private details.
- Edit the draft until it sounds like you. Save the edited prompt in a note on your phone.
- Use the checks. Read every draft before it goes out.
- After two weeks, see which prompts you reused. Drop the rest.
If you want help turning drafts like these into pages that bring in calls, that’s part of our SEO and content marketing work. For more on how AI answers pick local businesses, see how to get recommended by ChatGPT.
Sources
- U.S. Census Bureau · May 26, 2026Which firms use AI? (Business Trends and Outlook Survey)
- U.S. Chamber of Commerce Technology Engagement Center, with Teneo Research · Oct 29, 2026Empowering Small Business: The Impact of Technology on U.S. Small Business, Small Business Report 2026
- OpenAI Help Center · Oct 29, 2026Data Controls FAQ
WebByline
WebByline is a web design and local SEO agency in Dublin, CA, working with local and trade businesses. Posts under this name are team posts and updated versions of our older articles.


