The State of Robotics & Autonomous Systems APIs

About this paper

This is the least publicly published market API Evangelist has scored, and the distinction in that sentence is the report. 546 robotics and autonomous systems companies, averaging 12.6 against a whole-catalog 22.9, with a median of 10.0 and no Exemplars at all. Seventy percent sit in the Minimal band. Twenty-six companies publish an OpenAPI. Contract quality is 4.1 and governance 2.4, each with 89.3% of the market at zero — the lowest facet readings in this series.

This market is not short of APIs. It is short of public ones. Almost every product in this cohort ships an interface — a vendor REST or gRPC API on the controller, a ROS graph inside the robot, MAVLink on the radio link, OPC UA or MTConnect on the plant network. What almost none of them does is publish it where anyone outside the customer relationship can find it. The most likely reason is security posture, and the perception of it: a machine that moves has a physical blast radius, and the instinct — reinforced by customers’ security teams, safety obligations and insurers — is that a documented, discoverable interface is an attack surface. That is defensible, and it is not the same as having no interface. It does have a consequence: a gated interface cannot be discovered, evaluated or driven by anything the vendor has not onboarded, including the agents this market’s thesis depends on.

The report gives the honest structural defence in full: a robot’s control loop has no business running over HTTP, and ROS 2, DDS, OPC UA, MTConnect and MAVLink are excellent standards for the layer where they live. But the operational layer above them — fleet state, task assignment, telemetry history, mission upload — is where integrators work, where customers connect, and where an agent would act. It has no standard and almost no published interface, and that is the gap between the physical-AI thesis and the physical-AI interface, measured across 546 companies.

What's inside

1. Executive summary — the least machine-readable market in the catalog
2. How this cohort was built, and why it matters
3. The state of the market — distance from the hardware predicts the score
4. The resource taxonomy — too little published to have one
5. The composite Kin Score, and why it's the foundation
6. Agent-readiness — the gap between embodied AI and its interface
7. The blueprint
8. The standards layer — deeply standardised, where APIs cannot reach
9. The regulatory layer — where the rule requires data, the data appears
10. Standards and artifacts for the agentic turn
11. Security, auth, scopes and compliance
12. Provider-by-provider intelligence
13. The investable thesis
14. Where this is going

What you get for $500.00