Duck-Works Blog

Real world evals and physical autoresearch


The Arm Farm: a sandbox for embodied AI NEW!
Nov 13, 2026
Can Agents Learn To Drive?
Nov 12, 2026
Claude Stacks A Cube
Nov 4, 2026
Claude Construction: Multi-Agent Coordination
Nov 2, 2026

About this page

We evaluate AI agents in the physical world and create sandboxes for physical autoresearch.


Last updated: November 13, 2026