Blake Ruwali AI for the trades
AI for the trades
AI tools for roofers: calls, photos and claim paperwork
AI tools for roofers: how to handle storm-surge calls, pilot photo or drone-image estimates, draft insurance paperwork with a human check, and follow up.
AI tools for roofers help most with the busy-season office work: answering a flood of calls after a storm, following up on estimates, and drafting paperwork for insurance claims. Estimating from photos or drone images is worth a pilot, with a person checking every one. The inspection, the measurements that matter and the final price stay with your crew and your estimator.
Full disclosure: I run an HVAC company, not a roofing company. I haven't built any of this for roofing, so I'm not going to tell you how to read a roof. I can tell you what I've learned running a trades shop with AI, because the office problems are the same. The broader view is in what AI actually does for the trades.
The storm-surge problem
After a hailstorm or a windy week, the phone doesn't ring, it floods. Your office can answer a handful of calls at once. Everyone else gets voicemail, and a homeowner with water in the attic doesn't leave a voicemail. They call the next name.
An HVAC shop has the same pattern in a heat wave or a cold snap. At Home+ we use an AI phone agent for calls the office can't pick up. It collects the problem, the address and a callback number, asks a few questions, and puts the job into our system for a person to follow up. It does not diagnose, quote prices or promise arrival times. The setup is in how to answer every missed call with AI and a checklist is at /missed-call-fix/.
For a roofer I'd shape the script around:
- Is water coming in right now?
- Is there storm damage, and is it safe to be near?
- Is a tarp needed before an inspection can happen?
- Has the homeowner started a claim?
The AI collects answers and hands off. Anything about safety, such as a damaged structure or exposed electrical, goes to a person fast.
Estimating from photos or drone images: a pilot
Our estimate copilot reads a tech's photos and drafts a proposal from our own price book. A person checks every draft. It never invents a price, because it only chooses from lines we already set. The details are in AI estimates with a human check.
Could that approach work for roofing, with ground photos or drone images? Possibly. I haven't done it, so run it as a small test:
- Choose one common job type.
- Have your estimator do the estimate the normal way and also capture the photos or drone images.
- Let the tool draft line items from your price book.
- Compare the two and write down every difference and why.
- After a couple dozen jobs, decide whether the drafts save time or just create cleanup.
An image can hide a soft deck, a flashing problem or a layer count. An estimator who has been on the roof decides what the job needs and what it costs. Treat the tool's output as a draft, never as the quote.
Insurance-claim paperwork
Claims mean a lot of writing: descriptions of damage, scope summaries, photo captions, cover letters and follow-up notes. AI is decent at turning your facts into clean, organized text.
My rules for it:
- Feed it only facts you have: your inspection notes, your photo list, the measurements your estimator took. Tell it to say "not noted" where something is missing.
- Remove customer names, addresses, policy numbers and claim numbers before pasting into a general chatbot. Add them back in your own document.
- A person reads every line. A claim is a record, and it affects what a homeowner gets paid and what you get paid.
- Don't ask it what a policy covers or whether a claim will be approved. That's for the insurer and the people handling the claim.
The writing is the part AI is good at. The decisions are not.
Follow-up
Roof estimates are big decisions, so customers think about them for a while, and then they get busy. The estimate dies in silence. A message on day 2 and day 7 in your own voice, that stops the moment the customer replies, is a quiet win. I describe the cadence in AI follow-up for unsold estimates. When a customer asks a real question, a person answers.
What I'd skip
- Letting AI quote a roof from a satellite image and send it out unchecked.
- Letting AI promise a customer that insurance will pay.
- Pricing from a general chatbot. It doesn't know your material costs or labor rate.
- Buying a package before you've counted how many calls you missed last storm.
Where I'd start
- Count the calls you missed in your last busy week.
- Set up missed-call handling with a short script and strict limits.
- Add one follow-up rule for unsold estimates.
- Try the photo estimate pilot only after your price book is clean.
When you talk to vendors, use the questions in how I'd pick an AI partner. Ask them to show a real call or estimate becoming a real record in your system.
FAQ
What AI tools can roofers use? Phone agents that answer when your office can't, follow-up messages for unsold estimates, and writing help for claim paperwork and customer messages. Photo or drone-based estimating is worth a small pilot with a person checking every draft.
Can AI write a roofing estimate from drone images? It can draft one from images and your price book, but I haven't built that for roofing, so I'd pilot it. Your estimator should review every draft because images can miss damage underneath.
Can AI help with insurance claim paperwork? It can draft descriptions, summaries and photo captions from the facts you give it. A person must check every line, and AI shouldn't decide what a policy covers or promise an outcome.
How do I handle the call volume after a storm? Send overflow calls to an AI phone agent that collects the problem, address and callback number, and flags urgent or unsafe situations for a person. Then work the list in order.
Will AI replace estimators or crews? No. It takes phone and paperwork load off the office. Inspecting a roof and doing the work stay with your people.