5 min read
This is an illustrative composite story, built from common situations solo trades owners describe. Names and details are invented; the workflow is real and repeatable.
It is 7:40 on a Tuesday night. Marcus has just finished cleaning brushes in his garage when his phone buzzes again. Another text from a homeowner: four photos of a peeling hallway, a ceiling stain, and the question every solo painter dreads at the end of a long day. “Roughly how much would this be?”
Marcus runs a one-person interior painting business. He is good at the work and terrible at the paperwork. Before he changed anything, a typical week looked like this: eight to ten quote requests by text, two or three evening drives to look at rooms in person, and estimates typed on his phone at the kitchen table. He was spending around six unpaid hours a week on quotes, and losing jobs to whoever replied first.
Here is how he fixed it with tools he already had, and what he learned along the way.
The real problem was not pricing, it was speed
Marcus knew how to price a job. He had years of experience in his head: square footage, number of coats, how much prep a water stain needs, whether trim and ceilings were included. What he could not do was turn that knowledge into a clean, written estimate within an hour of a request coming in.
Painting is a crowded trade. The US Bureau of Labor Statistics notes that a large share of painters are self-employed, which means homeowners usually have several options a text away. The first clear, professional answer often wins.
Step one: put his pricing brain on paper
Before touching AI, Marcus spent one Sunday afternoon writing down how he actually priced jobs. Not a perfect rate card, just honest rules of thumb:
- A base rate per square foot of wall, by number of coats
- Add-ons for ceilings, trim, doors and accent walls
- Prep surcharges for stains, patching, peeling and high ceilings
- A minimum job price, so tiny touch-ups were still worth the drive
- What was never included (moving heavy furniture, wallpaper removal)
This one page turned out to be the most valuable part of the whole project. It also helped him later when he started moving toward flat project pricing, an approach we covered in how to move from hourly to project pricing.
Step two: let ChatGPT read the photos
ChatGPT has been able to look at photos since OpenAI added image and voice features, and Marcus already paid for a Plus plan. He created a ChatGPT Project (a folder that keeps instructions and files together) called “Estimates,” uploaded his pricing page, and gave it these standing instructions:
- Look at the photos and describe what you see: room type, surfaces, visible damage, ceiling height clues.
- List what you cannot tell from the photos and need to ask the customer.
- Draft a price range using my pricing rules, showing your assumptions.
- Write a short, friendly reply to the customer with the range, the assumptions and the open questions.
- Never present a range as a final price.
Now when a request came in, he forwarded the photos to the project, added a voice note with anything the customer had mentioned, and had a draft in about two minutes.
What actually happened in the first month
The honest results were good but not magic.
What worked: reply time dropped from “tonight or tomorrow” to under an hour for most requests. The drafts caught details he sometimes missed when tired, like asking whether the ceiling stain was from an old leak or an active one. He estimated he got back four of his six weekly hours.
What did not: the AI was poor at judging scale from photos. A narrow hallway shot from a low angle looked like a large room. Twice in the first two weeks it overestimated square footage by nearly half. Marcus fixed this by adding one line to his reply template: “Could you share the rough room size, or snap a photo with a door in the frame for scale?”
He also found the AI sometimes softened his minimum price when the job looked small. He made the rule stricter in his instructions, and he always read every number before sending.
Step three: the in-person visit became a confirmation, not a discovery
The biggest change was in how he used his evenings. Instead of driving out to look at every room, he only visited jobs where the customer had accepted the range. The visit became a quick confirmation of measurements and a handshake, not a sales call. His evening site visits dropped from about three a week to one.
After each job, he added a two-line note to the project: the estimated range, the final price, and what surprised him. Over time, the drafts got closer to his real numbers because the project had a growing record of his past jobs to compare against.
What Marcus would tell another solo tradesperson
- Write your pricing rules down first. The AI is only repeating your judgment faster. If the rules live only in your head, it cannot help.
- Quote a range, never a final number, from photos. Customers accept a range happily when the assumptions are clear.
- Ask for scale. One extra sentence fixed most of the measurement mistakes.
- Keep the final read human. Two minutes of checking beats one awkward conversation about a wrong price.
If quote requests come to you by phone rather than text, pair this with a way to capture calls you miss on a job site; our comparison of AI receptionists, voicemail and answering services is a good place to start.
What is the one piece of paperwork you keep doing at the kitchen table at night? That is probably your first project.
