What the system actually took in
The demo starts with 27 job site images from a real water loss. The system accepts up to 500 photos plus video, and it takes the job several ways: direct upload, a CompanyCam link, or a Drive link. You choose the job type, and if you want the project details pre-filled when the file opens in Xactimate, you enter them on the same form.
The price list is left blank on purpose. Per the video, you do not need it, because the estimate gets repriced inside Xactimate on import. Almost everything else on the intake form is framed the same way: not required, but every detail you add pushes the output closer to precise.
Dimensions are the one thing the field must supply
The system needs manual length, width, and ceiling height for each room. That is the single required input beyond the photos. The video teases a photos-to-sketch feature meant to replace DocuSketch, measurements straight from cell phone pictures, but at demo time the tape measure still rides along on the job.
Each photo takes about 15 seconds to analyze and build out the story of the loss, so how the room is photographed decides the ceiling here. The claimed precision, 95 to 98 percent on mitigation scopes and 92 to 93 percent on rebuild, is per the video, and it leans on that intake being complete.
What came back, and how it got priced
Detection finished in 193 seconds: 93 line items, 54 codes, six rooms built out as folders. Then comes a refinement phase for items the system, trained on thousands of real claims, is fairly sure exist but will not add on its own. The video is explicit about why: no phantom line items, ever. You can skip the questions and take the photo-only output, or answer them and push toward the higher precision number.
The download is a standard ESX. Import it into Xactimate, choose your price list and tax jurisdiction, and reprice. In the demo that produced a $12,253 mitigation estimate, which the builder says came in 98 percent precise against a scope he had written by hand for the same loss.
No phantom line items, ever.
Structure first, your judgment second
The video draws the line clearly. Precision is in the eye of the estimator: one writer scopes seven air movers where another writes six, and both can defend it. The system's job is 75 or 80 percent of the structure, and the estimator tailors the rest to the way their shop writes.
It also takes a shot at the market: companies claiming AI estimating while, per the video, routing your photos overseas to be written by hand and piped back into your system. The demo's counterargument is the clock. This pipeline became Axiom, and the review before the download is still where your estimator earns the file.
