Sandberg leads $10M round in Self Inspection for AI damage checks

The investor signal on this one is unusually loud. TechCrunch reports that Sheryl Sandberg, through her family office Sandberg Bernthal Venture Partners, led a $10 million round into Self Inspection, a San Diego-based startup founded in 2021 that lets users photograph vehicle damage on a smartphone and hands the assessment to a model trained on a large dataset of damaged cars. Jon McNeill's DVx Ventures backed the round alongside her, with strategic checks from tire distributor U.S. AutoForce and automotive lender Westlake Financial, plus Costanoa Ventures, Rebellion Ventures, and BrightCap Ventures.

The interesting part here is less the company and more the shape of damage-appraisal work in insurance, rental, and lending. Self Inspection's pitch is that its software guides the photo capture, compares the images against its damaged-vehicle dataset, and returns severity, cost estimates, and a detailed repair report, a workflow that has traditionally required a human adjuster to be somewhere near the car. According to the startup, it has processed more than one million vehicle inspections and claims to have cut customer costs by over $80 million and saved more than 300,000 operational hours. Take those numbers as reported by the company, not independently audited.

The customer list is the more useful proof point. Stellantis' financial services division reportedly uses the platform for corporate vehicle and lease-end evaluations, which is exactly the high-volume, low-margin workflow where a good-enough model beats a scheduled adjuster visit. Self Inspection says it will use the money to build additional products, expand its enterprise customer base, and enter the European market.

The honest caveat is that the reporting doesn't tell you what the current accuracy claim looks like, how the company prices its inspections, or which European geographies it is aiming at first, and Europe brings its own regulatory drag on both data handling and insurance workflows. What it does tell you is that a very high-profile operator is willing to bet that smartphone-based computer vision has crossed the threshold where lenders and fleet operators can start pulling the human out of the loop, and that the winners in this category are more likely to be defined by which enterprise contracts they lock in first than by any narrow lead on model quality.