Kyle Thomas

Kyle Thomas spent his RAF career asking a deceptively simple question: when can you trust a machine to act on its own?

Trained at the Empire Test Pilots’ School as an Evaluator Pilot, he worked on advanced air systems in environments where testing was not theoretical. The job was to understand what technology could do, where its limits were and what happened when real-world conditions did not match the assumptions made by its designers.

He later brought that experience into the RAF’s Rapid Capabilities Office, directing programmes in collaborative autonomy and edge intelligence and working on AI-driven air-system research. The technology was advancing quickly. The harder problem was proving that increasingly autonomous systems could be trusted when the consequences of failure were real.

In 2023, Thomas left the RAF and co-founded SAIF Autonomy, now Andorite Labs.

The company is building the trust layer for physical AI – software designed to safeguard autonomous systems operating in safety and mission-critical environments across air, land, sea and space. Rather than building the autonomous machine itself, Andorite focuses on the infrastructure that helps determine whether its behaviour can be trusted.

That distinction matters as autonomy moves beyond controlled environments. Drones, robots, aircraft and spacecraft increasingly need to make decisions without constant human intervention. Thomas’s argument is that intelligence alone is not enough – those systems also need independent safeguards around what they are allowed to do.

It is a problem he first encountered from the other side of the cockpit.

He spent years testing where autonomous systems reach their limits. Now he is building the layer designed to keep them inside them.