A credit committee cannot inspect every robot. It can inspect their record. RoboRun's underwriting rests on three questions, asked of the contract, the asset and the fleet's own telemetry.
The engines we built to annotate, validate and translate robot data are what let us answer them for a fleet in the field.
Which platform, how many units, what they cost, and what they would be worth if returned. Manufacturer support, spares and the depth of secondary demand all shape the residual value we are willing to lend against.
Who the customer is, what they have committed to pay, for how long, and what lets them stop. Service-level terms, termination rights and how fees are measured decide how much of the contracted revenue we can treat as reliable.
What the fleet has actually done. Uptime, task completion, failures and recovery, utilization and location, drawn from the robots' own telemetry and checked against source. This is where RoboRun differs from a conventional lender.
A conventional facility asks for a monthly compliance certificate. A monitored fleet reports continuously. Units that fall outside the agreed thresholds are flagged, reviewed against source evidence, and held out of the borrowing base until they are cleared.
Nothing in the underwriting stack is new. It is the annotation, validation and translation work RoboRun already does for training data, pointed at a lender's questions instead of a model's.
Video, state, actions and telemetry are aligned once, so every later claim about a robot resolves to a moment and its evidence.
Actions, outcomes, failures and recovery are labeled per episode. For a lender, these become task completion and incident rates per unit.
Deterministic gates clear most of the record; the rest goes to review. Units whose data fails the gates are held out of the base.
Any robot stack's telemetry maps into one covenant schema, so every capital partner reads every fleet the same way, and a facility can be refinanced without rebuilding the data.
We will tell you early and plainly whether a fleet is something we can look at. Most of the work happens on data you already have.
A week of telemetry from a few units, the contract or PO, and the fleet's cost basis.
We run the sample through the pipeline and come back with what we see, what we'd need, and whether a structure is plausible.
Counterparty, contract and asset review alongside a fuller fleet record. Covenant thresholds are set from the data, not from a template.
Documentation, first draw, and a live connection from the fleet to the facility's reporting.
Send a sample from a few units and the agreement behind them. We'll tell you what we see.