How to Write a Contractor Estimate with ChatGPT (Free Prompts)
ChatGPT cannot price your job — it does not know your market’s labor rates, your supplier costs, or your margin, and it will happily make up numbers if you let it. What it is genuinely good at: turning rough job notes into a clean, client-ready estimate description in under a minute, and building good/better/best options once you hand it real numbers. Below are three copy-paste prompts and a real worked example. You still supply the price — AI just writes it up so it reads like a $10,000 job instead of a text message.
Every contractor has the same estimate problem: the job is clear in your head, but by the time you type it into an email at 9pm it comes out as three run-on sentences and a number. Clients read a sloppy estimate as a preview of a sloppy job. AI does not fix your pricing — that is still on you — but it turns “redo the bathroom, tile and vanity, maybe 6 days” into a paragraph a homeowner forwards to their spouse without editing it first. That is the whole use case, and it is worth ten minutes to set up.
What AI actually gets you (and where it stops)
Be honest about the split before you start, or you will end up sending a client a made-up price:
- What it’s good at: turning your rough notes into professional prose, building tiered options, matching a tone, catching missing details you forgot to mention (permit? disposal? paint?), and doing it in seconds instead of twenty minutes.
- What it cannot do: know your local labor rate, know what your supplier charges you this week, know your margin target, or know if a client is a pain to work with and needs a padded number. Every price in your estimate has to come from you, not the model.
- The rule that keeps you safe: never let AI write a number into an estimate. Feed it your numbers and let it write the words around them.
The 3-prompt estimate system
Three prompts, used in order, cover almost every estimate you will write. Copy them into ChatGPT, Claude, or any chat AI — swap the bracketed parts for your job.
Prompt 1 — turn job notes into an estimate description
Prompt 2 — build good / better / best options
Prompt 3 — tighten the tone
A real example: bathroom remodel
Here is Prompt 1 run on the raw notes above, unedited except for formatting:
Input notes: “demo existing tile and vanity, install new 12×24 porcelain tile floor, tile tub surround, new 36” vanity, new toilet, repaint, approx 6 working days”
AI output: “This estimate covers a full renovation of your 5×8 bathroom. Work begins with careful demolition of the existing floor tile and vanity, followed by installation of new 12×24 porcelain tile across the floor and a matching tile surround around the tub. We’ll install a new 36-inch vanity and a new toilet, then finish with a full repaint of the space. Estimated timeline is 6 working days from start to final walkthrough. Materials, disposal, and daily cleanup are included in the scope below.”
Notice what did not change: no price appeared anywhere, and no scope item was added that was not in the original notes (no new lighting, no plumbing rough-in — nothing invented). That is the test for whether AI wrote this responsibly. Now you paste in your labor and material cost, add your markup, and the estimate is ready to send.
Where contractors get this wrong
- Letting the model invent scope. If you leave a detail vague, AI will often fill the gap with something plausible-sounding. Read every sentence against your own notes before you send it.
- Accepting a placeholder price. Some models will draft a “sample” number if you don’t explicitly tell them not to. Always instruct it to leave pricing out, and never paste in a number it suggested.
- Skipping the specifics. “New tile” reads worse than “12×24 porcelain tile, 40 sq ft.” Feed the model real counts and dimensions so the output has them too — specificity is what makes an estimate look pre-priced instead of guessed.
- No disclaimers on the estimate itself. AI-polished language is still an estimate, not a contract. Keep your standard validity window (“good for 30 days”) and change-order language on every estimate you send, written by you.
AI Prompt Pack for Contractors
25+ tested prompts covering estimates, proposals, scopes of work, and collections messages — the exact prompts in this guide plus dozens more, organized by job type so you’re never starting from a blank chat window.
Skip the prompting — start from a template
If you would rather not run three prompts every time, our estimate templates already have the client-ready structure built in — you fill in the scope and numbers and it is done. And if the job needs a full written scope of work rather than a short description, the free Scope of Work Generator builds that from a short form in about a minute, no prompting required.
Frequently asked questions
Can ChatGPT price my job for me?
No. It does not know your local labor rates, your supplier costs, or your margin. Use it to write up the scope and description clearly — you always supply the price.
Is it safe to send a client an AI-written estimate?
Yes, as long as every fact in it came from you. AI is writing the sentences, not deciding the scope or the price. Always read it against your original notes before sending.
Which AI tool should I use — ChatGPT, Claude, or something else?
Any of them work for this. The prompts in this guide run the same in ChatGPT, Claude, or Gemini — use whichever one you already have access to.
How do I stop AI from inventing scope items?
Tell it explicitly not to add anything beyond your notes, and always proofread the output against your original job notes before sending it to a client.
What is the fastest way to do this if I don’t want to write prompts?
Use a ready-made estimate template and fill in the blanks, or run the Scope of Work Generator for a full written scope in about a minute.
Get the Tuesday Template
One field-tested back-office template or AI prompt in your inbox each week. No fluff.