Video in, valuation out.
That's what we built for Chattel Valuations NZ.
Before
They came to us with a ChatGPT-assisted spreadsheet.
The expert valuer walked the room with a camera, typed line items into the spreadsheet afterwards, and mapped each one into the depreciation categories the report is built on. ChatGPT helped draft the prose around it. What it couldn't do was produce the report itself. A drafting assist isn't the same shape as a signed valuation.
What we built
There's a customer side and an engine side. On the customer side: a signed engagement letter, the jobsheet, the walkthrough video — the things the valuer would otherwise gather across a phone call and a site visit. On the engine side: it identifies what's in the room, estimates what the valuer would put on each item, runs the depreciation, drafts the report. The valuer still signs every one. The engine just does the line-by-line work that used to take days.
The engine keeps evolving as the work does — new categories, new edge cases, new outputs, without the workflow breaking.
Weeks into the engagement, the engine is already taking on edges we never scoped. That's the tell.
the tell.
The argument
Chattel Valuations is the first business we ran this loop on. It won't be the only one. We're after businesses where the bottleneck keeps moving — the engagement keeps moving with it, and so does the software.
Read how we partner, or talk to us.