How a Solo Chimney Sweep Turned Two-Minute Voice Notes Into Same-Day Inspection Reports (An Illustrative Story)

Weathered old brick chimney against a pale blue sky

6 min read

This is an illustrative story. Dana is a composite based on common challenges in one-person chimney service businesses, not a real individual, and the numbers are realistic examples rather than reported results.

It is 7:40 pm on a Tuesday in early November, and Dana is sitting in her truck outside her fourth job of the day, soot still on her sleeves, typing an inspection report on her phone with one thumb. The homeowner wants it before they light their first fire of the season. Two more customers from this morning are waiting for theirs. She has done this every night for three weeks, and she still has a full calendar until Christmas.

Dana runs a one-person chimney sweep and inspection business. October through December is her busiest stretch by far, because everyone wants their fireplace checked before the cold sets in. The sweeping itself is not the problem. The problem is the paperwork that follows every job.

Why the reports were eating her evenings

Every inspection ends with a written report. Customers expect it, insurers sometimes ask for it, and it is how Dana explains any repairs she recommends. Industry guidance, like the inspection levels described in the NFPA 211 standard and the homeowner resources from the Chimney Safety Institute of America, means her reports need to be specific: what she inspected, what she found, and what she recommends.

Her old routine looked like this:

  • Take photos and scribble notes on a clipboard during the job
  • Drive to the next job, forget half of the details
  • Spend 25 to 35 minutes per report at night, typing from memory and notes
  • Send reports a day or two late, which delayed repair approvals

With five or six jobs a day, that meant two to three hours of typing every evening, and repair quotes that sat in limbo because customers had not seen the report yet.

The one change: talk now, write later

Dana did not buy new software. She already had a free AI assistant on her phone (in her case, the free ChatGPT app), and she changed one habit: instead of writing notes, she recorded a two-minute voice memo before leaving each house.

Standing in the driveway, she would talk through the job the way she would explain it to a colleague: “Masonry chimney, single flue, light creosote, crown has a hairline crack on the north side, cap is missing, damper works fine, recommend a cap and crown sealing, no urgent hazards.” It took about the same time as writing on the clipboard, but it captured far more detail while it was still fresh.

Building a report template the AI could follow

Next, she spent one Sunday afternoon setting up the part that made it work. She wrote a report template with fixed sections: property details, what was inspected, findings, photos referenced, recommendations, and urgency. Then she wrote short instructions for the AI:

  • Use the template’s headings exactly, in the same order
  • Write in plain, calm language a homeowner can understand
  • Never add a finding that is not in the voice note
  • Mark anything uncertain with “[Dana to confirm]” instead of guessing
  • Keep the recommendations section as a short numbered list

That fourth rule mattered most. An AI that fills gaps with confident-sounding guesses is a liability in a safety report. Telling it to flag uncertainty, instead of hiding it, turned the AI from a risk into a reliable first-draft writer.

The nightly routine, now

At the end of the day, Dana opens each voice memo, pastes the transcript into the AI with her template and instructions, and gets a clean draft in seconds. She reads it, fixes anything marked “[Dana to confirm]”, attaches her photos, and sends it. Each report now takes her about six to eight minutes instead of half an hour.

Keeping customer details out of the AI

Dana was cautious about one thing: putting customers’ home addresses and names into an AI tool. Her fix was simple. In the voice note she only says “job four” and the chimney details, never the address or the homeowner’s name. She adds those herself in the final document. She also checked the data settings in her AI app and turned off the option that lets her conversations be used to improve the model.

If you handle sensitive client details, our comparison of local AI versus cloud AI for client work explains when it is worth keeping drafts on your own device instead.

What changed after a month

In this illustrative example, the results over the first month looked like this:

  • Report time dropped from about 30 minutes to about 7 minutes per job, saving roughly two hours every working evening
  • Reports went out the same day instead of one or two days later
  • Repair approvals came faster, because customers read the recommendation while the visit was still fresh in their minds
  • Reports became more consistent, since every one followed the same structure

The faster reports had a side effect Dana had not expected. Same-day reports with a clear numbered list of recommendations made it easy for homeowners to say yes to a cap installation or crown repair on the spot. The paperwork that used to be a chore became the most persuasive part of her sales process.

What did not work at first

It was not perfect from day one. Her first few drafts read like a brochure, full of phrases like “ensuring peace of mind for your family.” She added a line to her instructions (“no marketing language, write like a tradesperson explaining facts”) and the tone settled down. She also learned to say measurements clearly in the voice note, because a mumbled “three-eighths” once came back as “three to eight.”

And she still reads every word before it goes out. The AI drafts; Dana signs. That is a rule she will not relax, because her name and her liability are on every report.

How to try this in your own trade

This approach works for almost any job that ends with a written summary: home inspectors, HVAC technicians, pest control, landscapers, mobile mechanics. To start:

  1. Write your report template with the sections you always use.
  2. Write five rules for the AI, including “never add anything not in my notes” and “flag anything uncertain.”
  3. Record a two-minute voice memo before you leave each job.
  4. Leave names and addresses out of what you paste into the AI; add them yourself at the end.
  5. Read every draft before sending. Always.

Try it for one week on your simplest jobs first. If the drafts are good, move it to the rest.

What is the one piece of paperwork that follows every job you do? Tell us in the comments, and we may turn it into a future template.

Related reading

Leave a Comment

Scroll to Top