Ex-Google DeepMind researchers launch $130M AI startup to build autonomous coding tools

Former Google DeepMind researchers have launched a new superintelligence venture with substantial backing from top-tier investors. Reflection AI’s $130 million funding and ambitious focus on autonomous programming tools positions it alongside other AI labs developing agent-based automation systems. The startup’s focus on practical programming tools represents an initial step toward their long-term vision of creating advanced AI capable of performing most computer-based work.

The big picture: Reflection AI has secured $130 million across two funding rounds, achieving a $555 million valuation led by prominent investors including Sequoia Capital, CRV, and Lightspeed Venture Partners.

• The startup raised an initial $25 million seed round led by Sequoia and CRV, followed by a $105 million Series A co-led by CRV and Lightspeed.
• Additional high-profile backers include Nvidia‘s venture capital arm, LinkedIn co-founder Reid Hoffman, and Scale AI CEO Alexandr Wang.

Key players: The company is helmed by two former Google DeepMind researchers with experience developing Google’s Gemini large language model series.

• CEO Misha Laskin previously helped develop the training workflow for Google’s Gemini language models.
• Co-founder Ioannis Antonoglou worked on Gemini’s post-training systems, which optimize language models after initial training to improve output quality.

Strategic focus: Reflection AI aims to develop autonomous programming tools as its initial step toward creating what it defines as superintelligence.

• The company will first build AI agents that automate specific programming tasks like scanning code for vulnerabilities, optimizing memory usage, and testing application reliability.
• Future plans include generating documentation for code snippets and helping manage customer application infrastructure.

Technical approach: According to job postings, Reflection AI’s technology strategy involves large-scale computing resources and multiple AI techniques.

• The company plans to power its software using large language models and reinforcement learning.
• Their development roadmap includes exploring novel AI system architectures.
• Training operations will utilize up to tens of thousands of graphics cards, indicating substantial computational requirements.