Skip to content
AIPressRoom

FundingFoundation Models & Infrastructure

Mirendil Closes $1B to Scale AI Agents

Mirendil closed $1B Series A led by Kleiner Perkins and Clearlake Capital to accelerate AI R&D with AI agents.

Behnam Neyshabur, CEO of Mirendil.

The big picture: San Francisco-based Mirendil closed a $1 billion Series A round led by Kleiner Perkins and Clearlake Capital. The startup, founded this summer, aims to democratize frontier AI R&D by employing AI agents to carry out advanced research and development tasks, with humans primarily making judgment calls.

Why it matters:

  • Radical Leverage: Mirendil’s approach promises to enable small teams to achieve R&D outcomes previously requiring large institutions, significantly lowering the barrier to entry for advanced AI research.
  • Agent-Led Development: The company’s own competitive coding model, Mirendil-Base, was largely developed by AI agents, demonstrating the potential for recursive self-improvement in AI R&D.
  • Democratizing Access: By reducing reliance on extensive human staffing for pretraining and experimentation, Mirendil aims to shift the primary constraints to compute, data, and human judgment.

How it works:

  • AI Agent Orchestration: Mirendil employs AI agents to autonomously design experiments, fix engineering problems, and perform hands-on R&D tasks, betting on recursive self-improvement.
  • Human-in-the-Loop: The system integrates human oversight primarily for critical judgment calls, optimizing the efficiency of the research process by offloading complex execution to AI.
  • Lor Platform: Mirendil’s internal platform, Lor, allows hundreds of agents to write code and run experiments for extended periods, serving as the operational backbone for its agent-led R&D.

The catch: The promise of recursive self-improvement (RSI) remains a significant technical challenge, and Mirendil’s success hinges on its ability to consistently deliver on this ‘holy grail’ of AI. The nascent market for agent-led R&D tools could also face skepticism regarding the reliability and safety of autonomous AI systems in critical research contexts.

Key Facts

  • Company: Mirendil
  • Amount: $1B
  • Round: Series A
  • Investors: Kleiner Perkins and Clearlake Capital (co-lead), Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC, S32
  • Founders: Behnam Neyshabur, Harsh Mehta, Shayan Salehian, Tara Rezaei
  • Sector: AI Research & Development
  • Headquarters: San Francisco, US

AIPressRoom Weekly

Every round we reported this week, the people behind the companies, and the events coming up. One email, every Tuesday.

More in Foundation Models & Infrastructure

All