Brian Zhan’s Post

🚨 Excited to share 🚨CRV’s partnership with ReflectionAI from the Seed and Series A, and to join them on their journey. Founded by Misha Laskin, Ioannis Alexandros Antonoglou, and a team of former DeepMind and OpenAI pioneers behind ChatGPT, Gemini, AlphaGo, and AlphaZero, Reflection AI is tackling one of the most ambitious challenges in AI: superintelligence. Today’s LLMs are broad but often lack the depth to handle complex, multi-step tasks reliably. Reflection AI is changing that. Their autonomous coding agents—built on principles from AlphaGo and AlphaZero—can tackle well-scoped engineering tasks with remarkable precision. Solving coding agents will lead to superintelligence. From 2013-2020, the team created powerful narrow superintelligent systems with reinforcement learning: Deep Q Networks, AlphaGo, AlphaZero, and MuZero. From 2020-2024, they developed generally intelligent systems in the form of language models: PaLM, ChatGPT, GPT4, and Gemini. These projects shaped Reflection's strategy for how to build general superintelligence at Reflection: scale up the autonomous capabilities of language models with reinforcement learning. We are incredibly excited to partner with Misha and the team on this.