Direct answer. AI will not fix a broken restoration process. A new platform, chatbot, or automation cannot replace the drying tactic you already know, the channel that already produces work, or a three-minute job debrief written down while it is still true. Use the software you already pay for. Get what is in people’s heads into that system. Then the tool has something to pull from.
That is the argument I sat down with Phillip Rosebrook to make on Restoration Edge, the Business Mentors show for restoration owners. The episode title is the kit’s, and I am not rewriting it: AI Won’t Fix a Broken Process: Back to Basics for Restorers | Restoration Edge Ep. 8.
This piece is the operator version. Not a recap desk. The show is Phil’s. The claim is mine. If you want the conversation, watch it on the official player. If you want the move for Monday, stay here.
Key takeaways
- Tools do not dry buildings. Tactics do. An LGR, a desiccant, or a heater is equipment. Drying is the job.
- If a channel has worked for twenty-five years, do not kill it to “get on Facebook.” Dig the hole that already works.
- AI sitting in Encircle, Xactimate, HubSpot, or the Microsoft / Google stack you already pay for will beat a new bot with an empty brain.
- A job is a learning loop only if somebody records the debrief. Three minutes. Everybody talks. Do not bury the file.
- Scaling still starts with systems, not headcount — the same point as our scaling piece.
Watch the episode
Official video. Embedded from YouTube. Not downloaded, not re-uploaded.
Accessible summary of the full episode (25:59). Phil asks how restorers should think about AI and new software. Will’s through-line: do not lose the basics. Mindset, then tactics over tools, then keep what already works, then find the clients you can actually serve, then build systems before you add a model, then treat every job as a learning loop and run a real debrief.
What I told Phil about the next tool
Phil asked what a restorer should actually use if they are not building their own agent.
Not another platform. Whatever is already in the truck. Copilot if you are Microsoft. Gemini if you are Google. The model sitting in the CRM, the documentation tool, or the estimating software you already pay for.
The work is getting what is in your head into whatever that thing pulls from. Winter inventory. The client who never pays. The thing that always happens in January. It is not going to be you on every ticket. It is going to be the best version of you that shows up — if you wrote it down.
Accessible text. Phil frames the “what should I buy” question. Will’s answer is the opposite of a shopping list: pick the AI already sitting in Microsoft, Google, or the CRM, then feed it the knowledge that currently lives in one person’s head.
Tactics beat tools
It is almost never the tool. It is the tactic.
If all I have in the back of the truck is an LGR, or a desiccant, or just a heater, we still know how to dry. Fancy equipment is not the tactic. The tactic is drying. Same sentence in AR. Same sentence in AP. Same sentence in ops and sales. When you focus on the tool, you pigeonhole yourself.
Restorers should not lose the basics. A voice layer on documentation is an input, not the first purchase. IICRC protocols live in the workflow, not in the logo on the dehumidifier.
Start with what already works
I asked a shop what they were doing before they jumped on Facebook. Newspaper ads. Twenty-five years. Does it work? Yes. Then do not stop to go dig somewhere else.
Same rule on software. If Xactimate is how you get paid, the AI that matters is the one that helps that file survive the profitability dashboard. An estimate is what you ask to be paid. Job costing is what it cost you. Mixing those two up is a process failure. No model fixes that.
Stronger systems before AI
Tony Dungy won Super Bowls on blocking and tackling. One apostrophe in the wrong place and the bot sends you chasing the wrong thing. Slow down. Get the boring stuff. It probably starts with the owner. What are you doing?
Owners stuck at “me plus a van” usually do not have a lead problem. They have a them problem. That is the opening of Scaling a Restoration Company. Hire after the process is written. AI without a documented estimating process invents one faster, and with more confidence. AI governance is the same idea with a claim number attached.
Treat the bot like you would treat a human. Context. Tools. End result. I have had people tell me they spend eight hours a day in Claude. I do not have a measured sample size for that. If your stack is the product, that is a claim to test, not a badge.
Turn the job into a learning loop
Phil asked about job debriefs. I said we do not do them enough. That is the receipt.
Do it on Zoom if you have to. Record it. Give everybody three minutes. The PM. The tech. Contents. Good client or bad client. Good job or bad job. What we could have done better. Then do not bury the file.
Next time a salesperson brings an opportunity, somebody can say they do not pay for 120 days and that will break us. That knowledge lives in people’s heads right now. This is the same failure mode as The trucks leave. The record doesn’t. Bring the judgment and the packet to the QBR, or you are asking to be rebid.
I do not have a measured before-and-after on shops that started recording these. Until a shop shares a change in collection days, rework, or bid margin, the honest line is I do not have a number for this yet.
This week
- Name the tactic on one live file before you name a tool.
- Use the AI already inside software you pay for. Do not buy a second brain this week.
- If a channel still works, dig. Do not pause it for a new surface.
- Run one three-minute debrief on the next closed job. Record it. Store it.
- If you cannot brief a model the way you brief a person — context, tools, end result — you do not have a tactic yet.
Questions
Will AI fix a broken restoration process?
No. AI amplifies whatever you feed it. If estimating, debriefs, and collections are not written down, a model will guess — confidently. Fix the process, then let the software you already pay for pull from it.
What should a restoration company buy first for AI?
Nothing new, if the current stack already has a model. Copilot, Gemini, Encircle, Xactimate, or the CRM you already license. The first job is getting owner and crew knowledge into that system.
Why do tactics matter more than tools in water damage restoration?
Because drying is the tactic. An LGR, a desiccant, or a heater is a means. A crew that understands drying can still complete the job when the favorite machine is not on the truck.
How long should a restoration job debrief take?
Three minutes per person is enough if you do it while the job is still true. Record it. Store it. Use it the next time the same buyer or the same loss type shows up.
Who are Will Tygart, Tygart Media, and Restoration Intel?
Will Tygart is the founder of Tygart Media. Restoration Intel is Tygart Media’s publication for property-damage restoration professionals. This article is Tygart-owned reporting from his appearance on Restoration Edge.
Who is Phillip Rosebrook and what is Restoration Edge?
Phillip Rosebrook hosts Restoration Edge for Business Mentors. Episode 8 is the conversation this article is built from. Watch on YouTube or listen on Spotify.
Get the commercial system
The Vault has the QBR template and the ERA kit this debrief plugs into. One email. No sales call.