We are building the condition intelligence layer for collector vehicles because the institutions that move capital in this market deserve a defensible reference standard. The system has to be built by people who know the asset class deeply enough to define what a score should mean, paired with engineering capable of encoding that knowledge into infrastructure that compounds.
Brian has spent thirty years inside the collector vehicle market. He currently manages two of the highest-value private Porsche collections in the Bay Area and counts Fortune 100 CEOs, professional athletes, and technology executives among his clients. They write him checks specifically because the institutions they were supposed to be able to rely on (auction houses, factory restoration programs, authorized centers) failed them at the moment of transaction.
He was born into the market. His father was a Porsche independent and franchised Alfa Romeo dealer who raced Porsches professionally. His mother won an SCCA National Championship in 1975 in a 1973 Porsche 911 that Brian still owns. The institutional access, the technical reference library, the existing client roster, and the dealer license that anchor AutoAsset all pre-existed the company.
Three things have to be true simultaneously for the condition intelligence layer to be buildable now, and they all are.
Multimodal AI maturity. Video understanding, computer vision, and OCR of decades-old technical literature only became feasible at the accuracy and cost this requires in the last eighteen to twenty-four months. The technical lift that needed a twenty-person ML team three years ago is achievable for a small focused team today.
Institutional asset class recognition. Insurers, lenders, and family offices have only recognized collector vehicles as a legitimate asset class in the past five years. The institutional demand for the data infrastructure to support that recognition is just beginning.
Generational expert handoff. The cohort of expert inspectors who built the current subjective system is retiring. The knowledge is leaving the market. Capturing it into a structured corpus now is a closing window.
Five years from now, someone will own this layer. The question is whether it is built by people who know the asset class deeply enough to define the scoring parameters, or by a generic ML company learning the asset class on the way. The first wins because the dataset compounds.