Skild AI: deployment and in-context learning are robotics' hidden pillars
deepakpathak · x · 2026-09-12
In an article titled "The Hidden Pillar of Robotics Research is Deployment," Skild AI founder Deepak Pathak explains how the company reached $100M ARR within 10 months of its first commercial deployment.
- Learning the initial procedure is only part of the challenge; most robot foundation models can already do that
- The real difficulty is adapting when the customer's procedure changes
- In-context learning is the answer: communicate changes via another demonstration instead of restarting a data-collection and training project
- This dramatically cuts the engineering needed to keep complex deployments useful as operations evolve
Pathak argues real-world deployment itself is an underrated pillar of robotics research.
Related event: Skild AI hits $100M ARR just 10 months after first commercial deployment(34 posts)→
More from AGI Musings
- Researcher Slams Higher Ed's 'Abstinence-Only' Approach to AI Education — RexDouglass · 2026-09-12
- Richard Sutton: 'Too much intelligence is not one of the world's problems' — RichardSSutton · 2026-09-12
- Kelsey Piper: Don't lie to AIs in training — you'll just breed paranoid lie detectors — liminal_bardo · 2026-09-12
- GPT-6 Astra turns a prompt into a 3D biomechanical simulator, but healthcare needs more than demos — Olivier__OG · 2026-09-12
- Can We Structurally Detect a Terminal Self-Continuation Interest in Recursive Self-Improving AI? — coherence · 2026-09-12
- From pencils to ASI: China's cheap-replica model may scale to superintelligence — AaronBergman18 · 2026-09-12