"If we could snap our fingers and get a pile of data... we would solve general robotics right now."
Rohan Paul Twitter · Rohan Paul (@rohanpaul_ai) · 2026-06-29
Startup CyberOrigin's CyberCode platform is presented as addressing the primary bottleneck in physical AI — not model quality but data infrastructure — by making real human manipulation data searchable, synchronized, and traceable for training robot policies.
Appears in
Extraction
Topics: robotics-dataphysical-airobot-learningdata-infrastructurevision-language-action-models
Claims
- The primary bottleneck in physical AI and robotics is data infrastructure, not better model architectures.
- Robot policy training requires vision, motion, language, robot state, and sensor streams aligned on a single timeline, and misalignment causes models to learn incorrect behaviors.
- CyberCode transforms real human manipulation data into a searchable, inspectable, quality-checked, and evaluation-ready operating layer.
- Better data infrastructure can matter as much as better model architecture for manipulation policies, world models, and VLA models.
Key quotes
"If we could snap our fingers and get a pile of data... we would solve general robotics right now." — Figure CEO Brett Adcock
For manipulation policies, world models, and vision-language-action models, better data infrastructure can matter as much as better model architecture, because the model can only learn from the structure, coverage, timing, and quality the data system actually exposes.