AI for Remodeling Companies: What Actually Works in 2026 (From Someone Who Runs One)

I co-own CityLux, a general contracting company that does remodeling work, with my partner Scott Jamison. So when I tell you where AI actually helps a business like mine, I'm not writing from a marketing deck. I'm writing from a job site.

Most articles about "AI for contractors" are written by people who have never pulled a permit in their life. This one isn't. Here's what's true, what isn't, and what I'd tell another owner in my position.

Where it genuinely helped: a permit question in a county I didn't know

A while back I was bidding a job in a rural county I don't normally work in. I needed to know their septic tank permitting policy before I could even put a number together, and that's the kind of thing that used to mean a phone call, a hold queue, and a callback two days later if you were lucky.

I asked Claude instead. It gave me a clear, accurate answer on the county's septic permitting requirements, and it held up. That's the whole story, and it's not a small thing. In our business, the cost of not knowing a local requirement isn't an inconvenience. It's a bid that's wrong, a schedule that's wrong, a client relationship that starts on the wrong foot.

That's the pattern I'd point other owners toward: AI is excellent at surfacing information you'd otherwise have to chase down by phone. Permitting questions, code requirements, material specifications, anything where the answer exists somewhere but finding it costs you an afternoon.

Where it hasn't worked yet: getting a finished quote out the door

Here's the part most of these articles won't tell you, because most of them aren't written by someone actually running the automation.

I've been trying to build an automated pipeline that takes a job's details and produces a finished, ready-to-send quote for the client. It's not there yet. The final delivery step, the part where it has to actually assemble a clean, correct, client-ready document, keeps falling short. Not because AI can't write, but because a quote has to be exactly right, every time, with real numbers pulled from real data, and "close enough" isn't good enough when it's got your company's name on it and a client's money attached to it.

I say this because if I only told you the septic tank story, you'd think AI is magic. It isn't. It's very good at some things and genuinely unreliable at others, and the difference usually comes down to one thing: how well it's grounded in real, specific information versus how much it's guessing.

The actual lesson

The permit question worked because it was a factual lookup. The quote automation hasn't worked because it requires pulling exact numbers from my company's real data and getting them right every time, and general-purpose AI wasn't built to do that reliably.

That's the distinction I'd tell any contractor or remodeler to hold onto: AI that's answering from general knowledge is a great research assistant. AI that has to be right about your company's actual numbers, contracts, and policies needs to be grounded in your actual documents, not guessing from what's typical in the industry.

We're getting there on the quote problem. When we do, I'll write about it here, including what didn't work along the way.

What I'd actually recommend to another owner

If you run a trades or remodeling business and you're wondering where to start:

• Use it for research and lookups. Permit questions, code requirements, material specs. It's fast and, in my experience, accurate on this kind of factual question.
• Don't trust it yet for anything client-facing that has to be exactly right, like final quotes or contracts, unless it's working from your actual documents and numbers, not general knowledge.
• The gap between those two is the whole ballgame. That's true whether you're using a general tool or something built specifically for your business.

That second point is exactly the problem I started building Gable to solve, which is a longer story for another post.