Company profile
Fern
Fern builds learned world-model simulators and reinforcement-learning environments for robot-policy evaluation and improvement. Its platform fits action-co...
Facts
- Segment
- DEVELOPMENT_PLATFORM
- Country
- United States
- Headquarters
- San Francisco, California
- Founded
- 2025
- Website
- https://fern.bot
About Fern
Fern builds learned simulation and reinforcement-learning infrastructure for robot-policy evaluation. Its action-conditioned world models are trained from real robot data and let teams benchmark policies without repeatedly running physical hardware.
The platform targets scalable comparisons, failure analysis and policy improvement across checkpoints. Fidelity and transfer depend on the source data, robot embodiment, task distribution and model coverage, so simulated results still require physical validation.
Market context
ENABLEMENT
Summary
Type: RL infrastructure
Core: Learned simulators
Function: Policy evaluation
Audience: Robotics teams