Lake Stevens Pete (owner of an HVAC company and other businesses)
An owner of several small businesses, including an HVAC company, counted 27 active AI agents and 36 routines (26 on a schedule) in use across his companies. They pull statements, file invoices, update sales and cost sheets, flag overdue maintenance and track renewal dates. His rules: only three emails ever send on their own, everything else is a draft he approves, and nothing posts publicly or pays a bill without his yes. He keeps an inventory of all agents, which caught a failed reminder and two routines doing the same job.
On its own Jacksonville HQ renovation, design-builder Haskell posted QR codes around the jobsite so workers and trade partners could ask an AI agent about specs, submittals and drawings and get answers with links to the source documents. Assistant PMs also use a templated agent that compares a trade partner's submittal to the spec and lays out variances in a side-by-side table. Haskell estimates it saves about 10 minutes per field question and more than an hour per submittal review.
The roughly 20 plumbers at Oak Creek Plumbing & Remodeling near Milwaukee carry tablets with ChatGPT. They photograph a broken water heater or type up what they see, and use it to draft work proposals and invoices and to brainstorm fixes for tricky problems. President Dan Callies told CNN that even older techs have learned to ask it the right questions and that it has helped the company "lower our overhead and provide better services or better value."
New York City installer Mpower Solar uses AI-assisted design to build proposals for tricky canopy and tilt-mounted rooftops and adjust them live in meetings. Aurora reports proposal creation went from about 30 minutes to under 10, the change-order rate dropped from 50% to under 10%, and site surveys went from 2 hours to 30-60 minutes. The case study also says consistent design documentation led to fewer questions on submissions to state agencies and utilities.
Aurora SolarVendor case studyAurora Solar · Aug 2025
Remodeler Jimmy McKinney subscribed to OpenAI's service and trained his own version of ChatGPT on his company's past scopes of work for kitchen and bath projects. He gives it a written or spoken prompt about a new job and gets a draft SOW quickly, built on his own templates. Qualified Remodeler's editor says this one shortcut is already making the company more efficient.
ChatGPTTrade publicationQualified Remodeler · Mar 2025
Freedom Forever, which handles about 5,000 residential solar permits a month, built AI agents to read and act on the flood of utility and municipal permit emails (inspection notices, payment requirements, correction requests, approvals). The agent processed about 19,000 permit emails in its first month and fully handled about 71% without a human, at $0.20-$0.35 per run. A backfill turned up emails that had sat unprocessed for an average of three weeks, which got stalled projects moving again. A second agent validates uploaded utility bills, including photos of paper bills.
Claude Agent SDK, Claude CodeVendor case studyAnthropic (Claude)
Brett Tesson of Tesson Roofing & Exteriors in St. Louis uses EagleView Assess, an autonomous roof-inspection drone whose software uses AI to flag hail and wind damage and generates automated reports with measurements and 3D models. In a testimonial on EagleView's site, Tesson says the company had a $500,000 roof claim denied, and that after flying the Assess drone and giving the insurer the report, which proved what they were saying, the insurance company paid for the roof. The account is vendor-published; the claim details haven't been independently verified.