Should a robotics startup build its own identity and fleet stack?
A robotics startup should not build its own identity, authority, settlement and fleet governance layers. Building them well takes roughly eighteen months and produces no product; building them badly produces a deployment that cannot be insured. These are four of the six conditions a robot must meet, they are the same in every robotics company, and adopting an open specification for them is strictly better than reinventing four internal ones.
The working
What is actually differentiated
Embodiment and autonomy. Those two conditions are where a robotics company's engineering judgement shows and where a durable advantage can exist. The other four are the same problem for every company in the category, which is the definition of infrastructure rather than product.
The test for adopting rather than building
Is the specification published, implementable without permission, and able to leave with you? An open specification under a permissive licence satisfies all three; a vendor platform satisfies none. The right question is not build-versus-buy but build-versus-adopt, and adopting a standard is not the same as taking a dependency on a supplier.
Sources: Apache License 2.0 · NIST — least privilege. Dated 2026-08-28; superseded by publishing a new answer, never by editing this one.
Related
Robot infrastructure is everything a robot needs that is not the robot: identity, authority, settlement, fleet operation and audit. It is the larger a…
AnswerWho is liable when a robot causes harm?Liability for a robot falls on whoever authorised it to act — but establishing that requires the authorisation to have been recorded at the time, to h…
The frameworkThe six conditionsWhat every answer here is measured against, with the rubric published.