Blake Ruwali AI for the trades
AI for the trades
How I'd pick an AI partner for my trades business (questions to ask)
How to pick an AI partner for an HVAC, plumbing or electrical business: 15 questions to ask, red flags, what a fair pilot looks like and how to compare options.
To pick an AI partner for a trades business, choose the one that can show you real work in a real shop, connects to the systems you already use, starts with one small pilot, and lets a person approve anything that touches money or safety. Ask for a live demo on your own data, written limits on what the AI will do, a clear price structure and the ability to leave with your data. Here are the questions I'd ask, since I've been on both sides: I run a trades company and I build AI for others.
*Disclosure: I run Techeo, which builds AI for HVAC and plumbing shops. The questions below apply to us too, and you should ask us all of them.*
First: know what you need
"AI" covers different jobs. Write down your one biggest problem before talking to anyone:
- Calls going unanswered
- Estimates taking too long or going cold
- Follow-up that doesn't happen
- Marketing that doesn't turn into booked jobs
- Back-office paperwork
A partner who solves a problem you don't have is a distraction. Pick one problem and a measure you can count yourself, such as calls answered or estimates followed up, not a promise about revenue.
Questions about proof
- Do you run or work inside a trades business, or only sell to them? There's good work from both, but ask how they learn what actually happens on a service call.
- Can I see it working on a real call, estimate or job, not a slide deck? Ask for a live demo and a recording of a bad call, not only a good one.
- Can I talk to a current customer in my trade and size? A reference is stronger than a testimonial on a website.
- What does it get wrong, and how often do you find out? An honest answer names failure modes.
- What results are you willing to put in writing? Be wary of vague promises like "more revenue." Prefer commitments about process (responding to every lead, a review step) over outcomes no vendor controls.
Questions about fit
- Does it connect to my field service software (ServiceTitan, Housecall Pro, Jobber, or whatever I use)? Ask to see a record created in the system, not a notification.
- What data does it need from me, and who owns it? You should own your customer data and be able to export it.
- What does setup need from my team? Hours per week, who, for how long.
- How does it handle my price book and my rules? An AI that doesn't use your pricing is guessing.
Questions about control
- What can the AI say or do without a person? Get the limits in writing: no prices it can't verify, no diagnoses, clear safety handoffs.
- Who reviews its work, and how? Ask to see the review screen.
- How do I turn it off? A pause button you control, with no penalty.
- How is customer privacy handled? Where recordings and transcripts live, who can see them, and your obligations on call recording consent in your state.
Questions about money
- What's the price structure? Setup, monthly fee, usage charges, and what happens when volume spikes. Ask for a worst-case month.
- What's the exit? Contract length, notice, data export, and what you keep if you leave.
Red flags
- No live demo on real data
- "Guaranteed" revenue, rankings or savings
- Reluctance to explain how it works or what it does wrong
- Locking you into a long contract before a pilot
- No way to review or pause the AI
- Pressure to decide this week
- Reviews or case studies you can't verify
What a fair pilot looks like
A good pilot is 30 to 60 days, one workflow, one measure. Example: after-hours calls only, measured by how many produce a booked or scheduled follow-up record. Weekly review of recordings. A written agreement on what happens at the end: expand, change or stop. If the partner resists a pilot, ask why.
How to compare options
Make a simple scorecard with columns for each option and rows for proof, integration, controls, price, exit and how it felt to work with them. Score each 1 to 5 after the demo, not before. Include "do nothing" and "use ChatGPT and a checklist" as options, because sometimes that's enough. I explain where plain ChatGPT stops in ChatGPT for contractors.
Build, buy or hire
- Buy a tool when the problem is common (answering calls) and you want to be live fast.
- Hire a partner when you need it connected to your systems and rules, and you don't have time to run it.
- Build in-house only if you have someone who'll own it long term.
My own rule
I only trust AI that I'd be comfortable having my own customers use. That's why every Techeo tool starts at Home+ Air & Heat. Ask any partner what they run on their own business.
If you'd like a printable version of these questions, comment AI on my videos or message me.
FAQ
How do I choose an AI partner for my HVAC business? Pick one problem, ask for a live demo on real data, check it connects to your software, and start with a 30 to 60 day pilot with clear limits and an exit.
What should I ask an AI vendor before signing? Ask what it does wrong, what it can say without a person, who owns the data, how pricing scales and how you leave.
How much does AI cost for a small trades company? It varies by tool. Costs usually combine setup, a monthly fee and usage charges, so ask for a worst-case month in writing instead of relying on a headline price.
Should I hire a consultant or buy software? Software works for common, well-defined jobs. A consultant or partner helps when the AI must connect to your systems, price book and processes.
Is a long contract a red flag? Often, before a pilot. Start short, measure one thing, then decide.