Uptime is the robotics product

Generalist policies are improving quickly, and that improvement is largely irrelevant to the purchase decision. The buyer has a labour line, a throughput target, and a tolerance for stoppage. A robot enters the budget when it beats that line — including the people who keep it running.

Every robotics company therefore becomes an operations company with a hardware line item, whether or not the founding team wanted that business.

The honest model is payback period, and it must sit above the line: hardware cost, integration, downtime, spares, and remote-operator load. A warehouse fleet we tracked hit its contracted pick accuracy and still needed a remote operator roughly every twenty interventions. The customer's payback was eighteen months. The vendor's, at that intervention rate, never arrived.

The company survived by halving the intervention rate on its second hardware generation and by pricing remote operations explicitly instead of absorbing it. That is the whole business model in one sentence.

Autonomy claims are unfalsifiable until someone fixes the denominator. We use interventions per hundred task completions, sourced from operators rather than from vendor demos, because it is the number that converts directly into staffing cost.

Two systems can both claim high autonomy and differ by a factor of five on this measure. Only one of them is a product.

Structured environments already have cheap automation — fixed sensors, conveyors, jigs. Unstructured ones punish reliability gaps brutally. The band where a mobile generalist beats both is real, and thinner than the decks assume.

Our inspection case file is instructive: the robot was slower per site than a two-person crew, cheaper once travel was included, and approved on the safety line rather than the cost line. Expansion then stalled the moment sites were structured enough for fixed sensors.

The second-generation bill of materials tells you whether the company can live in that band. If gen-two cost does not fall enough to support the price point the model assumes, the fleet stays a pilot regardless of how good the policy gets.