Field Notes · · 12 min read

What an HVAC Owner Actually Needs to Know About AI

Skipping the hype. Three real jobs an HVAC company in Windsor used AI for last month — what worked, what flopped, and what the phone-answering bot cost them.

Source material Field report from Local Nerds install
AI in Crayon · ep
Three jobs. One bot. Real numbers.

This isn’t theory. This is a client report.

A Windsor HVAC company, 14 techs, 3 office staff, implemented three AI workflows last month. Here’s what actually happened.

Job 1: Call triage bot

The goal: Pick up after-hours calls, capture lead info, book the actual-emergency calls into the on-call rotation.

What worked: Capture rate went from ~40% (voicemail) to ~85% (bot took the info). Two emergencies got properly routed.

What flopped: The bot misread one caller’s accent and booked the wrong address. Tech drove 40 minutes the wrong way. $180 wasted.

Job 2: Estimate write-up

The goal: Tech dictates from the truck, AI cleans it into a customer-ready estimate doc.

Result: Works. Estimate turnaround went from next-day to same-day. Closed-won rate up slightly.

Job 3: Service history summaries

The goal: Before a maintenance visit, hand the tech a one-pager on the customer’s last three visits.

Result: Techs loved it. Customers noticed. One asked if they’d hired a “customer intelligence person.”

The Monday test

Don’t try all three. Pick the one where a bad outcome costs you the least, and run it for 30 days. Measure. Then decide.

Keep going

AI in Crayon · ep
Fifty jobs → one judgment
Up next
Borrowed Lenses

Fifty Specializations Collapse Into One Skill

What happens to a 15-person service company when the things you used to hire five specialists for become one person with good judgment? The math is better than you think.