Research note 002
Physical AI: the gap between a demonstration and a deployment
Robotics and autonomous systems need an evidence standard that goes beyond a compelling video. Track the operating environment, the customer and the repeatability of the result.
A deployment needs an operating context
Our structured tracker contains 97 deployment records as of 2026-10-29. This is coverage collected by TLDRocket, not a census of robots in operation. A record entering the tracker is not proof that a customer has deployed a system at scale.
Physical AI introduces questions that a software launch can leave unanswered. A machine operates in a particular environment, with a defined task, safety constraints and human support. A successful demonstration in one setting does not establish the same performance in a different warehouse, factory or road network.
Follow the evidence through three stages
Demonstration: a system performs a task under stated conditions. Capture who produced the evidence and whether unsuccessful attempts are visible.
Pilot: a named customer tests the system in an operational setting. Record duration, supervision and the success criteria agreed before the test.
Production: the system performs a recurring business function. Look for customer confirmation, an operating period and measured outcomes against a baseline.
These are research categories, not automatic endorsements. Where a source does not establish a stage, keep the stage unknown. This avoids turning vendor publicity into a procurement recommendation.
The detail behind
the decision.
Read the complete analysis, use the decision framework and bring the findings into your workspace.
Explore intelligence plans ↗Already a member? Sign in →