Today, we are announcing Proximal.

We are making a single technical bet: The majority of improvements in coding agent capabilities will come from increasing the quality and scale of reinforcement learning environments. Results in mathematics and competitive programming have demonstrated that models can outperform the world’s best humans given sufficient training environments. Our job is to enable the same results in software engineering.

Doing this requires solving hard engineering and research problems:

How do build sandboxes that let models to work with external dependencies such as databases and cloud infrastructure?

How can we measure code quality beyond functional correctness?

How do we build agents that inspect our environments to automatically uncover edge cases and detect reward hacking?

Our founding team has worked on RL infrastructure and coding agents at companies like Cursor and Prime Intellect and has previously founded and successfully sold companies.

We are backed by top investors and angels from OpenAI, Anthropic, Thinking Machines, xAI, Meta Superintelligence, Cursor and Cognition.