AI for Restoration Companies: 4 Things You Can Do This Week With ChatGPT or Claude

RIA Staff
RIA Staff
on Thu, 11/19/2026
AI for Restoration Companies: 4 Things You Can Do This Week With ChatGPT or Claude

Insights by Harris Bergman

If you run a small restoration shop with no IT department, a basic ChatGPT or Claude subscription can make a big impact on your day to day.

1. Turn Messy Field Notes Into Clean Job Documentation

Your teams already know what happened on the job but getting it out of their heads and into the file is hard. AI tools can streamline this process.

How to use it

  • Have techs open the company ChatGPT or Claude app after each visit and just talk. Rooms, materials, readings, equipment. Rambling is fine.
  • Paste the transcript in and ask for a structured daily log.
  • Upload job photos and ask for a room-by-room description of visible damage.
  • At the end of the job, paste in all your notes and ask for the loss story in date order.

Try this prompt

“You’re an experienced mitigation PM familiar with IICRC S500. Here are my notes from today’s water loss. Turn them into a daily job log with the date, rooms affected, actions taken, equipment placed, and moisture readings. Only use what’s in my notes. If something is missing, flag it instead of filling it in.”

2. Get a Second Set of Eyes on Your Estimate

AI won’t replace your estimator, but it’s a great gut check before you send it out the door.

How to use it

  • Export the estimate to PDF and upload it with a handful of job photos.
  • Ask which line items are commonly missed on that type of loss.
  • Upload the estimate and the final invoice together and ask where they don’t match.

Try this prompt

“You’re an experienced restoration estimator familiar with IICRC S500. This is an estimate for a Category 2 water loss in a finished basement. Based on the photos and scope, list line items that are commonly missed on jobs like this and explain why each one might apply. Don’t assume anything the photos or scope don’t support.”

3. Write the Emails You Keep Putting Off

Every contractor has a list of emails they mean to send. Like that 40-message thread nobody wants to open again that’s waiting on your PM. These emails matter… AI can make it a little easier.

How to use it

  • Paste in a long email thread and ask for a short summary and a draft reply.
  • Build templates for your most common emails. Save them and reuse them.
  • Ask for a homeowner update written in plain language with no industry jargon.

Try this prompt

“You’re a restoration office manager who works with insurance adjusters every day. Write a polite but firm follow-up to an adjuster. Our invoice was submitted 45 days ago and we haven’t received a response. Ask for a payment status and a date. Don’t add any details I haven’t given you.”

4. Review the File Before It Goes Out

A job can be done right and still get held up because something is missing from the file… AI can catch a lot of it before the carrier does.

How to use it

  • Upload the daily logs, moisture readings, photos, and estimate for one job.
  • Ask it to flag missing readings, gaps in dates, or equipment days that don’t match the logs.
  • Ask which line items on the estimate don’t have a photo or note to back them up.
  • Give it your company’s job file checklist and ask where the file falls short.

Try this prompt

“You’re an experienced mitigation PM familiar with IICRC S500. Review this job file before it goes to the carrier. Check that every day of equipment has matching readings, every line item on the estimate is supported by a photo or note, and nothing on our checklist is missing. List each gap and where you found it. Don’t fill in or guess at missing information.”

Ground Rules Before You Start

AI is useful but it is not your estimator, your lawyer, or your project manager. It will make mistakes and it will make things up. Always check the results.

  • Keep customer data off personal accounts. If you can, buy AI at the enterprise level so your team is on the same tool. If confidential information is going into the tool, use a business plan that keeps your data safe.
  • Turn off model training. In ChatGPT it’s Settings > Data Controls > “Improve the model for everyone.” In Claude it’s the model improvement toggle under Settings > Privacy. 
  • Tell it not to make things up. Add “don’t fill in missing information” to any prompt that touches readings, line items, or job details.
  • Read everything before it goes to a carrier. AI can get facts wrong and sound very sure of itself.
  • Give it context. It doesn’t know your carrier guidelines, your local pricing, or how your company works unless you tell it.