Hello.

We’re Entrain, a team in Seattle building the inference layer for physical AI.

A chatbot can take a second to answer. A robot can’t. The world keeps moving while the model thinks, and an action that arrives late is a dropped part or a stopped line.

Robotics companies build the machines and train the models that drive them. The hard part has moved from training those models to running them, in real time, every time. That’s the part we work on, for every kind of robot.

For the teams who make robots, that means bigger models on the hardware they already ship. For the teams who put them to work, it means robots that run the whole shift without stopping to wait on their next move.

We’re still in stealth, so that’s all we’ll say here. If you’re building physical AI, we’d like to talk to you.