The claim is decided before the crate closes
Ask your underwriter where your losses come from and the answer is the road. AXA XL's fine art underwriters say most of the claims they see are transit or water damage. Storage and display barely register. The damage happens once an object is in someone else's hands, in a truck, through customs, across a handoff between custodians your registrar has never met.
The list of ways that goes wrong is longer than most people outside this work expect. Carriers writing inland marine cover for fine art underwrite against improper crating, shock and vibration, climate swings, water intrusion, and storage in uncontrolled conditions during a customs delay. Most of those leave no mark until someone opens the crate.
And when a piece arrives damaged, the argument is never really about the damage. It is about what you can prove about the condition it left in. The claim is settled by the record built beforehand, which is careful, unhurried work, and it is usually the first thing squeezed when a show has to ship on a date.
Everywhere else in logistics, the AI pitch is speed. In your world that pitch lands as slightly alarming. Nobody on your team wants a faster handler.
Your failure mode is an incomplete record. Nobody has ever lost a claim over a slow one.
What the technology already does today
At Yembo I built ScanMyHome. Walk a gallery with a phone and what comes back is measurement-grade: the room's dimensions, the wall cutouts, the doorway you have to get a crate through, and the objects in the space identified individually. No tripod, no laser, no specialist. Do the same in your storage and your vault is mapped. Do it at a venue before a traveling show and you know what fits before the truck is loaded.
A gallery is an interior space. So is a vault, a crate room, and the venue you are installing into next month. The geometry does not care what is hanging on the wall.
None of that is a roadmap item or a demo. ScanMyHome's output exports into Xactimate, which means insurance carriers already settle property claims against those measurements, and that is a useful bar for whether a number is trustworthy enough to argue over.
I am still not going to tell your registrars that software should write condition reports. It should not, and anyone selling you that has not handled anything fragile. A model that has seen ten million pieces of household furniture has never seen craquelure and has no business grading it. What it can do is guarantee that every object in a shipment was measured, photographed, catalogued, and timestamped, so your trained eyes spend their hours on the pieces that need judgment rather than the ones that only needed recording.
The practical guidance on insuring art on the move comes back to the same thing: document before it ships, because nothing you do afterward substitutes.
So what I bring your room is the view from inside a capability that already works on their spaces: what it does today, what it will be marketed as doing next year, and exactly where those two diverge. That is available about three years before the vendors selling into this market will admit any of it.
The other side of the crate
There is a second reason none of this is abstract to me. I make art. Lightwall is a kinetic light installation I built with the artist Rita Sus: glass, motors, radar, and a model that responds in real time to whoever walks up to it. It premiered at the California Center for the Arts Museum, and it is being prepared to exhibit at Dutch Design Week in Eindhoven in October 2026.
Which puts me, this year, on the other end of your paperwork. Getting an installation of glass and embedded electronics from a museum in California to a venue in the Netherlands is a fine art shipping problem, and being the person whose work is inside the crate teaches you something that reading claims data does not. The record I would want if that piece arrives wrong is the record I am arguing for on the rest of this page.
My work on AI in the moving industry covers the operational side of this in more depth, and my work on AI for insurance covers what happens on the carrier's side when a claim is contested.
