Our first AI product entered commercial use in 2020, at a point when automated damage assessment was still unusual in insurance. It has been in production every hail season since. What follows is what we can state publicly. Several customers do not permit naming, and we prefer an accurate anonymous statement to a vague named one.
Deployments and results
In commercial production since.
Countries with active licences.
Calculation matrices, twelve global and ten customer specific.
Interface languages.
Where it runs
Embedded in a global AI claims solution provider. A leading international provider of AI claims technology has integrated our detection and measurement module into a hardware assessment product sold under its own brand, together with performance engineering work carried out against their hardware.
Deployed by a top five Japanese insurance company. Rollout began in 2026 with the insurer's own claims operation, with extension to dealerships and partner networks planned.
Independent assessors and repair networks. Licences are active across more than twenty five countries, from single practitioners through to organised networks working on behalf of insurers.
How accuracy is measured
An accuracy figure without a method behind it is marketing. Ours is measured by comparing automated detection and sizing against a manual reference assessment performed on the same vehicles, counting dents per panel zone and per size band, which is the unit the claim is actually settled in.
Two things are worth stating plainly. Results depend on capture conditions more than on anything else, which is why the application constrains capture so tightly. And the honest test is not ours: we would rather scope a comparison against your own vehicles and your own reference assessments than ask you to accept a number produced in our conditions.
Standards
Output follows the calculation conventions used in the markets we operate in, rather than a private scheme of our own. This includes the BVAT calculation specifications and hail tables used across the German-speaking market, metric dent size range models, and coin based sizing as used in the United States and Canada. Where a customer works to its own matrix, that matrix is built and maintained as part of the deployment.
Security and data handling
Hosting regions, data retention, application security and the questions that arise in insurer procurement are covered separately. See security and compliance.
Test it against your own claims.
A bounded pilot on your vehicles, your matrix and your reference assessments, run alongside your existing process.