Refresh builds reinforcement learning environments that provide verifiable reward signals for training AI models, focusing on automating computer use and coding workflows.
Founded
2025
YC Batch
X25
Yc Primary Partner
Diana Hu
Product Features & Capabilities
Customizable reinforcement learning environments, verifiable reward signals for training AI models, sandbox environments for real-world task execution, focus on tasks capturing real economic value, designed for AI research labs and software engineers automating coding tasks.
