I resisted AI tools for about two years. My work is physical, I pull cable, mount hardware, configure networks. There's no prompt that does any of that. So when everyone started talking about AI productivity tools, I mostly tuned it out.
Then I started tracking where my unbillable time was actually going. Proposals: two hours each, three or four a week. Scheduling back-and-forth: 45 minutes a day. Follow-up emails: another hour. I was spending 15 hours a week on admin for a business built on labor. That's when I started paying attention.
1 · Why I was skeptical
My concern with AI tools for contractor work is the same concern I have with any automation: what looks like time savings often just moves the problem. Bad AI output means editing time, which erases the gain. I wasn't willing to send clients proposals that sounded like they were written by a chatbot. The bar was: it has to sound like me, or it doesn't ship.
2 · Tool 1: AI for proposals (ChatGPT)
I use ChatGPT to write the prose sections of client proposals, not the pricing, not the parts list, not the scope. Those I still build by hand because they require actual knowledge of what the job involves. But the introductory paragraph, the "why this solution" section, the closing, those are now AI-assisted.
My process: I write three bullet points describing the client's problem and my proposed solution, paste them into ChatGPT with the instruction "write this in the voice of a direct, experienced residential tech engineer," and edit the output down to something that sounds like me. Total time: 12 minutes per proposal instead of 45. I've sent 12 proposals this way and two clients commented that the write-up was unusually clear. That's the bar.
3 · Tool 2: Scheduling and follow-up (Calendly)
Calendly isn't new, but I added the follow-up sequence feature six months ago and it's the thing that actually pays for itself. When someone books an initial consult, they automatically get a confirmation, a 24-hour reminder, and a post-meeting follow-up asking if they have questions. I configured those three emails once. They now go out without me touching them.
For home tech work specifically, the post-meeting follow-up is where clients often remember they had another question. Having it go out automatically means I get that question in writing instead of a phone call two days later.
4 · Tool 3: Meeting notes (Otter.ai)
I started recording client walkthrough conversations with Otter.ai running on my phone. It transcribes in real time and produces a summary at the end. For a 45-minute site assessment, I get a five-bullet summary of what the client said they needed, what I said I'd look into, and any follow-up items. I paste that into the proposal file immediately after the meeting, while it's accurate.
The transcription isn't perfect, it mangles technical terms, but the structure is there and the key decisions are captured. It's replaced a notepad I was inconsistently keeping.
5 · What doesn't work
AI for pricing and parts estimation is useless for my work. Equipment prices change weekly. A quote for a UniFi setup I did in January is stale by March. No AI tool knows current distributor pricing, lead times, or which products I can actually source locally. That part of the estimate still requires looking up real quotes from real suppliers. Don't let anyone sell you an AI estimating tool that claims otherwise.