About SimReal
- Founded
- 2026
- Headquarters
- Shenzhen, China
- Team size
- 1–10 people
- Company focus
- A commercial company specializing in RL environments
- Products
- RL environments · Verifiers · Expert trajectories
- Task domains
- Enterprise workflows · Long-horizon tasks · Computer use
- Focus
- Enterprise workflows across Chinese and other non-US software ecosystems
- Open-source roadmap
- Publish reproducible verifier experiments first, then environment and data specifications.
- Model-company shareholders
- None
- Contact
- business@simreal.co
What we are building
SimReal builds reinforcement learning environments, invariant-based verifiers, and expert trajectories for frontier AI teams. Our stateful simulations let agents practice in work scenarios that can be reset and checked, preserving every action, decision rationale, and correction. We focus on education, marketing, office collaboration, mathematics, programming, and customer support, with a roadmap into Chinese and other non-US enterprise software ecosystems.
Project resources
- Month-end close: verifier attack experiment ↗Source code and results available
- Six task families · 18 scenario templates ↗Open the workspace
- Expert trajectory data specification ↗Evolving with the collection protocol