Reflection AI Raises $130 Million from Sequoia, CRV, and Lightspeed to Build Autonomous Coding Agents

Reflection AI, a New York-based startup founded in 2024 by former DeepMind researchers Misha Laskin and Ioannis Antonoglou, has raised US $130 million in early-stage funding. The funding comprises a US $25 million seed round led by Sequoia Capital and CRV, followed by a US $105 million Series A co-led by CRV and Lightspeed Venture Partners. Additional investors include the venture arm of ‎Nvidia Ventures, Reid Hoffman, Alexandr Wang, SV Angel and Lachy Groom.

Reflection AI’s mission centers on building fully autonomous coding agents and more broadly “superintelligent” AI systems—tools designed not just to assist human engineers but to read, write, test and deploy software independently. The founding team draws on previous achievements at DeepMind and Alphabet, with Laskin and Antonoglou shifting focus from behind-the-scenes reinforcement-learning work on ‌Gemini and AlphaGo into the frontier of autonomous engineering. The $130 million raise values Reflection AI at approximately US $555 million.

The company intends to deploy the capital toward scaling its engineering infrastructure, hiring top talent across research and product, and commercialising its autonomous-coding platform. Early commentary indicates the firm already has paying customers in industries with large engineering teams (such as fintech and enterprise software), and its agents are designed to integrate into corporate code-bases, project tools and communication workflows to tackle “well-scoped engineering tasks” end-to-end.

Investors view the move into autonomous coding as a landmark opportunity. Sequoia noted that while coding assistants increased developer productivity ten-fold, Reflection’s agents propose a leap beyond—placing entire development flows under autonomous control. CRV emphasised that the platform “has potential to redefine industries”, while Lightspeed highlighted the startup’s Reinforcement Learning + LLM hybrid as a leap toward general superintelligence. Nvidia’s involvement underscores the infrastructure-intensive nature of the work, especially given the scale of computing required to train such agents.

Despite the impressive funding and backing, Reflection AI faces notable headwinds. Delivering autonomous agents that reliably handle production-grade engineering tasks remains unproven at scale. The company must validate its technology, integrate deeply into enterprise workflows, ensure robust code-quality, security and version-control, and navigate the long sales cycles typical of enterprise-software adoption. In addition, building a commercial model for agentic systems—and proving ROI—will be critical as the startup moves from research to deployment.

Furthermore, Reflection AI operates in a competitive environment: the race for autonomous-software tools and next-generation agents is crowded with well-funded rivals from large tech companies and emerging labs. The company’s bold vision—to treat software engineering as a superseded human activity by autonomous agents—carries technical, organisational and market-adoption risk. The infrastructure cost alone for training frontier models is enormous, and competing for talent and compute remains costly.

With the funding secured, Reflection AI is positioned to execute on its vision. The capital gives it runway to build out infrastructure, expand its product offerings, and deepen enterprise engagements. As it seeks to transition from stealth to scale, milestones to watch include enterprise-customer deployments, revenue growth, as well as any public release of its autonomous-agent platform. If Reflection AI succeeds, it may redefine how software is built—and mark a pivotal moment in the evolution of AI from assistant-mode to autonomous-agent-mode.