SkyeBrowse is best known for videogrammetry: a pilot flies a drone over a scene, uploads the video, and gets back a 3D model. Its main customers have been public-safety agencies. On Thursday, Nov. 5, the company released Roof Damage Analysis, a feature built for a different market. It takes the same one-flight 3D model, finds the roof, measures it, screens it for storm damage and pulls in the local severe-weather history. The result is a report written for roofing contractors and insurance adjusters.
The software makes a judgment about damage, so the most important design choice is who has the final say. SkyeBrowse leaves it with the inspector. "The AI is a screening aid: you approve or reject every finding," the company wrote in its announcement. Nothing the model flags reaches the final PDF until a person approves it.
From video to facets
The workflow starts the way any SkyeBrowse job does. The pilot flies the property and uploads the video, and the platform builds a 3D model. Roof Damage Analysis then identifies the roof and outlines its facets, the separate planes that make up the roof surface. According to DroneDJ's coverage, if the software picks the wrong building, the user can click the correct one. Inspectors can drag facet corners and add or remove facets when the automatic outline is wrong.
From that outline the tool produces the measurements roofers price jobs from:
- Roof squares and total area
- Main pitch and facet count
- Ridge, hip, valley, eave, rake and step lengths
- A suggested waste percentage
SkyeBrowse published two demo results, reported by DroneXL. A commercial building came out at 17 facets, 4,665 square feet (433 square meters), 46.7 squares, an 8/12 main pitch, 249 feet of eave and 17 percent suggested waste. A house came out at 13 facets, 3,573 square feet, 35.7 squares and 18 percent waste. These are the vendor's own demonstration numbers, not results checked against hand measurements. They do show the level of detail the report is meant to deliver.
The 20-second damage screen
The damage scan is the AI part of the product. DroneXL reports that it reviews up to 16 frames from the flight video and up to 8 gallery photos in about 20 seconds. The list of conditions it looks for is long. DroneDJ lists hail impacts, wind creases, missing or lifted shingles, granule loss, exposed underlayment, cracked tiles, blistering, ponding, membrane tears, rust, dents in soft metal, debris and tarps. Several of these, such as ponding, membrane tears and rust, apply mainly to low-slope commercial roofs and metal parts rather than asphalt shingles. That suggests the tool is aimed at commercial roofs as well as houses.
The scan does not give a yes-or-no verdict. Each finding is pinned to a spot on the 3D model and labeled either Likely or Possible. The inspector then works through the pins. Each one is either approved, which puts it in the report, or rejected, which drops it.
SkyeBrowse also says what the scan does not do. Per DroneXL, it does not determine the cause of damage, whether a policy covers it or what repairs are needed. Those three questions are where most roof claim disputes happen. DroneDJ describes the tool the same way, as a screening aid and not a replacement for a professional inspection.
Storm history, with NOAA's own caveats
The third part is a weather tab. It lists hail, wind and tornado reports from NOAA's Storm Prediction Center within 25 miles (40 km) of the property, covering the year before the flight. The tool also pre-fills a date of loss from the nearest hail report. For an adjuster, that date is often the first field on the claim form.
That convenience comes with limits SkyeBrowse passes along. DroneXL notes that the SPC says its report list "likely does not contain all severe weather reports," and that NOAA labels the reports preliminary. According to DroneXL, the SPC compiles the list from local storm reports sent out by National Weather Service offices. A hailstorm that nobody reported will not appear, and the nearest recorded report is not always the storm that damaged a given roof. The pre-filled date is a starting point, not a finding. Like the damage pins, the inspector has to check it.
What ends up in the report
The output is a PDF with per-facet measurements, a roof diagram, photo evidence, the storm history, the methodology and a reviewer sign-off. That sign-off is how the human review shows up in the document. A reader of the report can see that a person reviewed each finding, not only that software flagged it.
The mapping side of the release
The same Nov. 5 update included features aimed at SkyeBrowse's existing users. A new 2D map download produces files ready to load into specific tools:
- SARTopo/CalTopo: a Web Mercator GeoTIFF
- ArcGIS Pro: a UTM GeoTIFF
- Google Earth: a KMZ
- AutoCAD/Civil 3D: a UTM GeoTIFF with a world file and a .prj projection file
A separate Contours DXF export adds elevation contour lines for CAD work. Support for ground control points (GCPs) lets users move a model onto surveyed coordinates. That matters whenever a drone model has to line up with survey-grade data and not just look right.
The SARTopo/CalTopo format reflects SkyeBrowse's public-safety roots. Those tools are widely used in search-and-rescue planning, and SkyeBrowse's own announcement says a search team can have the map in SARTopo minutes after landing. The ArcGIS and AutoCAD formats extend the same map download to GIS and engineering workflows.
Price and packaging
DroneXL reports that Roof Damage Analysis runs on SkyeBrowse Pro, listed at $49.99 per model or $39.99 on annual billing. It is also part of the Public Safety & Enterprise plan. SkyeBrowse overhauled its pricing in September, so this feature arrives under a pricing structure that is only about two months old.
Why It Matters
The release is notable for two reasons.
The first is the market move. DroneXL describes the product as aimed at roofing contractors and insurance adjusters, outside SkyeBrowse's public-safety base. A company whose videogrammetry has been used mostly at crash and crime scenes now wants to sell roof inspections. Claims and repair estimates depend on the numbers, so the measurements carry weight in that market. The underlying technology is the same: one flight in, a measurable 3D model out. Collision reconstruction and roof measurement both depend on whether that model is geometrically accurate. Adding GCP survey correction in the same update fits that emphasis on accuracy, though the sources do not tie it to the roof tool.
The second is how the AI is handled. In roofing and insurance, the important question is not whether a model can find a hail hit. It is who stands behind the finding when a claim is contested. SkyeBrowse's approach is to have the model screen, an inspector decide, and only approved pins reach the PDF. It also states openly that the system makes no call on cause, coverage or repair scope. That limits what the product claims, and arguably the company's exposure if a finding is wrong. It also puts responsibility on the inspector. If the screen is fast and usually right, reviewers will be tempted to approve pins without looking closely. The Likely/Possible labels and the incomplete SPC data are reminders that the software's output still needs checking.
Whether the screen holds up on real storm-damaged roofs, beyond the vendor's demo buildings, will depend on independent field use. None of the three sources cited here reports independent accuracy data for the measurements or the damage detection, so those claims are unverified for now.
