Blog

July 2026

Calibration Precedes Evaluation

How we calibrate real-world robot evaluations with an adaptive difficulty staircase to learn more, spend less, and iterate faster.

July 2026

Evaluation Precedes Capabilities

Integral to robotics development will be the art of evaluation: measuring where we are today, where we want to go, and the gap between those.

August 2025

Monosemantic Features in Vision-Language-Action Models

We trained transcoders on π₀-FAST to uncover interpretable features in robotics models, revealing two classes: "scene" features that highlight objectives and "ego" features focused on the robot itself.