What Fern does
Fern, from Ishiki Labs, builds simulation and evaluation technology for robotics, supporting what it calls physical AI at scale. The company creates high-fidelity, custom robot world models that let robotics teams evaluate and train robot control policies in detailed digital simulations before deploying them on real hardware.
Key capabilities
- Custom, high-fidelity robot world models tailored to specific environments
- Simulation environments for testing and training robot policies
- Evaluation tooling to refine AI models before real-world deployment
Who it's for
Fern targets robotics teams and organizations developing autonomous robots that need to evaluate and improve their AI policies efficiently. Its differentiation is an emphasis on accurate, application-specific simulations rather than generic environments, enabling more realistic policy evaluation. The company is Y Combinator-backed and has published a whitepaper and case studies on its approach.