Key takeaways
- Start with the work that leaks hours or jobs: the night call, the job file and the homeowner updates during a dry-out.
- Run any new AI use in shadow mode first, with a person approving every draft, before it talks to a customer on its own.
- AI must never change a moisture reading or a photo, quote what insurance will pay, or give health advice about mold or smoke.
- Measure before you start, or you will not know whether it worked.
The way to use AI in a restoration company is to start where hours and jobs leak: answering the night call, turning a technician's notes and moisture readings into a clean job file, and keeping homeowners updated through a multi-day dry-out. Pick one, run it for 30 days with a person checking, measure it, then add the next.
Where AI pays off first for restoration companies
AI pays off first in three places, each tied to how a restoration job actually runs.
1. The night call. Water losses do not keep office hours, and a caller who reaches voicemail calls the next company. An AI receptionist can answer, ask your triage questions, and page the on-call technician for active water. This is the use with the most direct dollar value, because every answered loss is a job you would otherwise not hear about. The AI receptionist page for restoration companies covers it in detail.
2. The job file. A water job produces photos, moisture maps, daily readings, equipment counts and notes, and someone in the office turns all of it into a file the adjuster will accept. AI can draft the daily summaries and the job narrative from what the technician recorded, so the estimator reviews instead of retyping. The readings and photos themselves stay exactly as the technician captured them.
3. Homeowner updates. A dry-out runs for days, and the homeowner wants to know when the equipment comes out, why the fans are so loud and what happens next. AI can draft those updates from the job record, answer routine questions, and send the review request when the job closes. Your coordinator stops fielding the same five questions all day.
Marketing drafts come later: service page outlines, ad variations, replies to reviews. They help, but they rarely pay as fast as the three above.
Adding AI one task at a time
- Pick one task One that repeats every day
- Set guardrails What it may and may not say
- Review its work A person checks before customers see it
- Measure the result Time saved, leads answered
Then add the next task
A 30-day starting plan
- Days 1 to 3: pick one use and measure it. Count last month's missed after-hours calls, or the office hours spent writing up each water job. That is your baseline.
- Days 4 to 7: write the rules. List what the AI may say or draft, what it must hand to a person, and the exact wording for safety situations such as a gas smell or water near the electrical panel.
- Days 8 to 14: run it in shadow mode. The AI drafts every reply or summary; a person reads, corrects and sends. Keep a list of what it got wrong.
- Days 15 to 21: go live for routine cases. Let it handle the cases it got right in shadow mode. Anything unusual still goes to a person.
- Days 22 to 30: compare and decide. Put the new numbers next to the baseline, then keep, fix or drop it before you start a second use.
Risks and guardrails
The risks in restoration are claims, evidence and privacy, so the guardrails are specific:
- No insurance promises. AI never tells a homeowner what their policy covers or what a claim will pay. The adjuster decides that, and a wrong promise becomes your problem.
- No changes to evidence. Moisture readings, drying logs and job photos are the record a claim depends on. AI can summarize them; it must never edit, fill gaps in or enhance them.
- No health advice. Questions about mold exposure or smoke and soot go to a person, who points the homeowner to a medical professional.
- Private data stays in business tools. Claim numbers, policy details and photos inside people's homes belong in systems you have an agreement with, not pasted into a personal chatbot account.
- No fake reviews. AI can draft a reply to a review. It must never write reviews; the FTC's rule bans fake and AI-generated reviews (16 CFR Part 465).
- Outbound AI calls need consent. Calls using an artificial or prerecorded voice fall under the TCPA's consent rules (47 CFR 64.1200), so use AI voices for answering, not cold outreach.
- Say it is AI. Callers and texters should know, and should always be able to reach a person.
Measuring it
Measure each use against the baseline you took in the first week, in numbers your office already tracks.
| Use | Before | After 30 days |
|---|---|---|
| Night call answering | Missed calls after hours; minutes to call back | Calls answered; active losses dispatched; jobs booked from night calls |
| Job file drafting | Office hours per water job file; days from job close to file sent | The same two numbers, plus adjuster requests for corrections |
| Homeowner updates | "When is the equipment coming out?" calls per job | Those calls per job; review requests sent and reviews received |
If a use does not move its number in 30 days, fix the rules or drop it. The AI tools guide for restoration companies compares the tools themselves.
How VETS Growth Marketing helps
We find the two or three places AI pays off fastest in your restoration office, build them into the software you already run, and keep a person in the loop where claims, evidence or safety are involved. Each one is a project with a date and a baseline, the accounts stay in your name, and we report the hours saved and jobs captured. Bring your busiest week to a growth consultation and we will point to where AI would help first.
Sources
- 16 CFR Part 465: Use of consumer reviews and testimonials, Electronic Code of Federal Regulations
- 47 CFR 64.1200: Delivery restrictions (TCPA rules), Electronic Code of Federal Regulations