“You Lost Me at AI” — Why That Reaction Is About to Cost Trade Owners

By Rob Heller

I got an email this morning from a plumber in North Carolina. I’d written something about how Swivl’s AI features were helping companies handle calls and estimates without throwing more bodies at the problem. His reply was four words.

“You lost me at AI.”

I have a lot of sympathy for that. I was running a plumbing company eight years ago, and if someone had told me an AI was going to answer my phone, I would have laughed at them too. The trades have heard a thousand pitches about technology that was going to change everything, and most of those promises evaporated the moment someone tried to use them on a Tuesday morning between three jobs.

But here’s the thing — and this is why I’m writing this post instead of just replying to his email. The skepticism is healthy. The opt-out is a mistake. The companies figuring out which AI tools are actually useful right now (and which are still hype) are going to put the holdouts out of business inside the next three years. Not because AI is magic. Because the tools that are real are quietly removing the most expensive part of running a service business — the humans you currently have to hire to handle volume that doesn’t actually need a human.

Let me walk through what I mean.

The Receptionist You Don’t Have to Train, Pay, or Replace

Bryan Landreth runs North Alabama Plumbing. He came on our podcast earlier this year and said something that stuck with me:

“An AI receptionist will know everything about your company because it’ll memorize basically your website. So it can speak intelligently about everything that you do as long as your website has got it in there. And it can book appointments for you, right, directly in there.”

— Bryan Landreth, North Alabama Plumbing

▶ Watch the full conversation: Bryan Landreth on Swivl Sessions, Episode 25

That used to be a $40,000-a-year job. Now it’s roughly a $300-a-month subscription that doesn’t take lunch breaks, doesn’t quit, and doesn’t fumble the call when three people ring at once on a Monday morning. It books the job, captures the address, asks the right qualifying questions, and drops the result into your dispatch board.

▶ Want to see exactly how to set this up in Swivl? Guide to Automating Calls & Setting Up AI Receptionist in Swivl — 4 minutes, screen recording, no fluff.

Ben Alexander at TRADEWORX put it this way on a different episode:

“Like a receptionist with AI — and you pre-programming it. ‘Here’s my top 50 questions’ — and they can actually book that on your calendar. There’s some amazing technologies coming up that are really going to revolutionize the home services industry.”

— Ben Alexander, TRADEWORX Consulting

▶ Watch the full conversation: Ben Alexander on Swivl Sessions, Episode 20

Ben is right. The operators who’ll win are the ones who already have it running.

The Estimator That Doesn’t Have a Bad Day

The second one is estimates, and this is where I see the most resistance, because estimating feels like a craft. The experienced tech walks the job, looks at the panel or the rooflines or the trench, and quotes from feel.

But here’s what’s actually happening in most shops: the estimate is being typed up by an office person who’s never been on a roof, working from a tech’s scribbled notes, in a spreadsheet that was last updated two years ago, with margins that haven’t been adjusted for material costs in eighteen months. That’s not craft. That’s a cost leak.

A trained AI estimator — and yes, I’ll name the one we built, Swivl Max — does the math the same way every time, pulls live material pricing, applies the margin you set on each line item, and hands the finished quote back to you in about two minutes. The tech still walks the job. The tech still makes the judgment call on scope. The AI does the part nobody enjoys and nobody does consistently at 4:45 PM on a Friday.

The Real Risk Isn’t Replacement. It’s Scale.

Most of the pushback I hear about AI in the trades is about jobs. “Are you trying to replace my CSR?” No. The jobs you actually want to keep are the ones where judgment, relationship, or hands-on skill matter — and AI is genuinely bad at all of those. What it’s good at is the high-volume, low-judgment stuff that’s eating your payroll.

The worry better worth your attention: your competitor across town just stopped paying a part-time receptionist, redirected that budget to two more vans, and is now answering 100% of inbound calls inside three rings. Meanwhile you’re sending half your calls to voicemail at 5 PM on Tuesday. That’s the gap AI opens up, and once it’s open it doesn’t close.

Ryan Hanson runs TRUE Plumbing — also on the podcast. He’s eyeing expansion from plumbing into HVAC and electrical, and the math on that kind of expansion only works if you’re stripping out linear hiring. Tools like AI receptionists are what let one operator credibly run multiple trades without doubling overhead:

“Adding HVAC and electrical is on the radar, but we’re still trying to figure out this marketing and smooth out the operations a little bit and have our team get a little more experience.”

— Ryan Hanson, TRUE Plumbing

▶ Watch the full conversation: Ryan Hanson on Swivl Sessions, Episode 29

Operators thinking like that are the ones to watch.

Where to Start Without Overcommitting

If you’re skeptical but you don’t want to get caught flat-footed, three suggestions:

• Start with calls, not estimating. The AI receptionist is the lowest-risk, highest-impact place to begin. Worst case you turn it off and go back to voicemail.
• Pick one trade or one job type first. Don’t AI-ify your whole operation in week one. Run it on residential service calls only, watch what happens for two weeks, then expand from there.
• Listen to fifteen recorded calls in the first week. Every AI call is logged and you can play any of them back. Inside seven days you’ll know whether to scale it up or pull it back. You cannot do that with a human receptionist.

You don’t have to be the early adopter. You just don’t want to be the last one.

The plumber in North Carolina who replied “you lost me at AI” — I respect the instinct, but I hope he reconsiders. The version of him that does is going to be running three more trucks in 2027. The version that doesn’t is going to be wondering where his calls went.

If you want to see what AI calls and AI estimates actually look like in practice — without having to fight a robot to find out — the free tier is here, and there’s a 2-minute overview of what Swivl does end-to-end on YouTube. Five minutes of poking around will tell you whether it’s nonsense or not. That’s all I’d ask.

— Rob