The Mit List | Episode 4: Joe on the Dollars in the Details
Joe Ledbetter has spent the last few weeks doing something most software people never do: sitting in a room with restoration crews, watching them try to do the thing his product is supposed to help with. One of the architects behind Encircle AI and Encircle Scope, Joe went into a couple of shops for hands-on training and came out with a finding that had nothing to do with photos, floor plans, or moisture readings — the stuff everyone assumes is the problem.
In this episode, Joe sits down with Leah, Encircle’s Director of Product and Content Marketing, to talk about what he actually found: crews are missing the narrative. Not the tasks. Not the readings. The story of the loss. And when he had techs role-play describing damage out loud, the results were good enough to give an estimator goosebumps.
TL;DR: Shops already collect photos, floor plans, moisture readings, and daily notes — and it’s still not enough to build a defensible scope. What’s missing is the actual description of the damage: what the space looks like, what’s wet, what it’s made of. Joe ran a role-play exercise with field techs and watched them go from a rough first attempt to “mind blown” on the second. The reason it matters: roughly 30% of a scope’s dollar value lives in details estimators normally have to guess at — containment, access, materials — and voice narration is what surfaces them. Encircle’s AI turns that narration into the scope. A one-click export into Xactimate is coming later this summer.
Joe Ledbetter
Principal Innovator – AI
Encircle
Leah Vusich
Director, Product Marketing
Encircle
Everyone’s collecting data. Nobody’s telling the story.
Ask any shop what they need for documentation and you’ll get the same list: photos, a floor plan, moisture readings, daily notes. That’s the checklist the industry has been trained on for years, and it’s not wrong. But when Joe sat in on in-field training with real crews, he found techs who were disciplined about all of it — removed baseboard, extracted water, set equipment, all logged — and still missing the one thing that actually explains what happened.
That’s the narrative. The plain-language description of the loss and the damage. Not the task list. The story.
“We’ve been fundamentally missing something, like as an entire industry. And that’s the narrative. That’s the actual description of the loss.”
Joe Ledbetter
Principal Innovator – AI, Encircle
The role-play nobody wanted to do — until they saw the output
Joe had techs stand in front of the group and describe, out loud, the damage in the room they were standing in. The first attempts were rough. First time doing it, no rhythm to it, plenty of stumbling. But when they saw what that rough description turned into — a scope built from their own words — the reaction wasn’t embarrassment. It was disbelief.
The second attempt was the real surprise. Techs went from beginner to expert-sounding overnight, and the scope itself stopped reading like a scope. It read like the best field note they’d ever produced — a narrative that actually described the loss instead of just checking boxes.
That’s the reframe Joe keeps coming back to: for the technician, this tool isn’t a scoping tool. It’s the note they’ve been asked to write for years and never quite pulled off. For the estimator or admin behind the desk, it’s still a scope. Same output, two different jobs finally getting what they each actually need.
Run this yourself: the exercise any shop can copy tomorrow
If you’re an owner watching this and want to try it with your own crew, Joe’s version doesn’t require much setup. Get your technicians in a room, open a new claim in Encircle, and have each person describe a fake loss out loud — give it a source, give it some context. No photos, no floor plans, just voice notes. Run the scope, compare outputs in real time, delete the note, and let the next person go.
The point isn’t the fake loss. It’s watching your own team see, in real time, what a good damage description turns into — and immediately want to do it better the second time, because now they know what “better” looks like.
What a D-grade damage description looks like next to an A-plus
The difference isn’t effort. It’s specificity. “The floor is wet, the walls are probably wet” gives you direction, but it’s generic — it could be any room in any house. “I’m standing in a kitchen on wood floors and they’re wet. There’s drywall installed on the walls and it’s wet” is the same observation with the environment described. That’s the gap between a D and an A-plus scope, and it’s entirely about how much context makes it into the recording.
Joe’s framing for techs who feel awkward narrating out loud: pretend FaceTime doesn’t exist and you have to explain the whole scene to your boss over the phone. That conversation — wood floors, cabinets, “we’re going to need more equipment” — is the damage description. It’s not a performance. It’s the same thing techs already do informally, just captured instead of lost.
The dollars are in the details — literally about 30% of them
This is where the exercise stopped being a training nicety and started being a business case. Joe’s team has looked at where estimates typically fall short, and the answer is depressingly specific: roughly 30% of most estimates are made up of the small stuff — containment poles, zippers for zip containment, the details an estimator often assumes happened but has no way to confirm. That assumption gap is exactly what a rich narrative closes.
Run the math on a $10,000 claim: without the narrative, you might land billing around $8,000 to $9,000 — full of holes, none of it defensible if a carrier pushes back. With the narrative, that missing $3,000 has a paper trail. Same job, same work performed, very different ability to collect for it.
That defensibility question isn’t incidental — it’s the same standard IICRC S500 has always pushed shops toward: documentation thorough enough to justify the scope of work and the decisions behind it, not just a record that work happened. A narrative-first field note is a more complete way to meet that bar than a checklist ever was.
Where estimators actually land on this — and where AI actually fits
When Joe walked these role-play scopes over to the estimators, the first reaction was relief: “I could write an estimate on this immediately — I don’t even have any questions.” The narrative was filling the gap estimators normally fill by guessing — flooring type, wall material, cabinet type, all the context that used to live only in the tech’s head.
There was a beat of nervousness, too — estimators wondering if the tool was coming for their job. Joe’s answer: no. The tool doesn’t replace the estimator, it hands them a complete file instead of a partial one. AI’s job in all of this isn’t to walk the loss — it can’t see, smell, or understand a structure the way a technician can. Its job is to do the heavy lifting on the data the technician already collected: turn a good damage description into the field’s best note, and turn that note into a scope the estimator can actually use.
What’s next: one click from scope to Xactimate estimate
The chain Joe describes — a two-minute voice narration becomes a note, the note becomes a scope, the scope becomes an estimate — is about to get one step shorter. Encircle’s estimating feature, matching the scope directly into Xactimate with a single button, is coming later this summer.
The sequencing matters. An estimate is only as good as the scope behind it, and the scope is only as good as the damage description that fed it. None of the automation downstream works if the field capture upstream is thin. Which is why Joe’s closing challenge is deliberately simple: on your next loss, open a general note in Encircle, hit voice-to-text, and just start talking — awkward or not. Then press generate scope.
Frequently Asked Questions
Not photos, floor plans, or moisture readings — those were already being collected. What was missing was the narrative: an actual plain-language description of the damage and the loss. Techs were logging tasks (removed baseboard, extracted water, set equipment) without ever describing what the space looked like or what had actually happened to it.
Techs took turns describing, out loud, the damage in the room they were standing in — no photos or floor plans, just a spoken description. The first attempt was rough. The second attempt, after seeing what the first one produced, was dramatically better — techs went from a beginner-sounding note to an expert-sounding scope in one repetition.
Gather technicians in one room, open a new claim, and have each person narrate a fake loss out loud — give it a source and some basic context, no photos or floor plans involved. Run the scope after each person, compare the outputs in real time, then delete the note and let the next person go. Seeing the comparison live is what teaches the team what a strong damage description sounds like.
Specificity about the environment, not just the damage. “The floor is wet” is generic. “I’m standing in a kitchen on wood floors, and they’re wet — there’s drywall on the walls and it’s wet too” describes the actual space. The second version gives an estimator (or an AI) enough context to work from; the first one doesn’t.
Because roughly 30% of most estimates consist of details — containment setup, access constraints, material types — that estimators typically have to assume rather than confirm. On a $10,000 claim, that’s about $3,000 in scope that either has documentation behind it or doesn’t. Without the narrative, that money is the first thing that gets challenged or cut during a carrier review.
No. The technician stays the domain expert — they’re the only one who can see, smell, and understand the structure in person, and AI can’t do that part. AI’s role is to process what the technician already captured: turning a spoken damage description into a complete field note, and that note into a scope the estimator can build from. Estimators still own the estimate; they’re just starting from a complete file instead of filling gaps by assumption.