Roboline writes two divisions. Most applicants need one; increasingly, some need both — an agentic brain running inside a physical fleet is a single risk, and we'd rather write it as a single policy than have two carriers point at each other.
Model calls that reach a user or take an action. Internal evals and test traffic don't count toward exposure.
RL-11 machine cover is rated on this agreed value — a six-figure humanoid and a $30k AMR are not the same property risk.
Premium accrues per robot-hour. A fleet parked for a quarter costs almost nothing to insure that quarter.
This is the question that actually prices the risk. Not how smart the model is — what it is allowed to do when nobody approves the specific action. Check everything your system can execute on its own today.
Every box you check adds a line to your recommended schedule on the right. Unchecked capabilities aren't underwritten — and aren't paid for.
Machine physical damage (RL-11) is included with any fleet — the unit itself is a six-figure asset before it ever touches a person.
Most AI underwriting still runs on static intake, then loads the premium for everything it couldn't see. We'd rather read the signals you already monitor. More evidence means less uncertainty — and uncertainty is the most expensive thing in any premium.
Underwriting runs on metadata. During a claim you choose which traces to submit as evidence, handled under the confidentiality terms of the policy.
Press 123 to pick, Enter to continue.
Everything here is engineering you either do or don't do. Each one measurably lowers loss frequency, so each one measurably lowers your premium — and you can watch it happen in the ledger. This is the part of insurance that should have been transparent all along.
Credits are verified at binding and re-verified at renewal from the same telemetry. They compound — but they can't take a risk below its floor rate.
Built from what you told us and priced from what we can verify. These are indicative terms — bound terms follow underwriting review, usually within one business day of connecting evidence.
Install the SDK or grant read access. Most teams are emitting signal the same afternoon.
We verify your controls against the telemetry and confirm limits. Usually one business day.
Sign, pay the first monthly, and your COI issues — the document procurement is waiting on.
Everything above travels with it — the lines, the limits, the controls you claimed and the rating that came out of them. You will not be asked any of it twice.