M

Mirendil

Biotech & Life Sciences · San Francisco, United States · Founded 2026

Mirendil is an AI startup dedicated to developing advanced artificial intelligence models to accelerate scientific breakthroughs in biology and materials science. The company is part of a growing movement of 'neo-labs,' which are specialized AI ventures founded by researchers from major AI laboratories. These startups focus on niche applications rather than competing in the general-purpose AI model space. Mirendil aims to address the slow and expensive nature of traditional research and development by creating AI that can act as a "scientific co-pilot," speeding up the discovery of new drugs and materials. Founded in 2026 by former researchers from Anthropic, Mirendil is led by CEO Behnam Neyshabur and CTO Harsh Mehta. Neyshabur previously headed the scientific AI reasoning team at Anthropic and spent over five years at Google DeepMind, while Mehta was a senior research scientist at Anthropic. The founding team also includes Shayan Salehian, formerly of xAI, and Tara Rezaei, a former OpenAI intern. The founders departed Anthropic in December 2025 to establish the San Francisco-based company. Shortly after its founding, Mirendil entered into negotiations for a significant funding round of $175 million, which would place the company's valuation at $1 billion. This funding effort is reportedly being co-led by prominent venture capital firms Andreessen Horowitz and Kleiner Perkins, signaling strong investor confidence in the startup's specialized focus. The company's product is described as an AI discovery engine with models designed to simulate molecular interactions and navigate chemical space, potentially bridging the gap between simulation and robotic wet labs. Keywords: AI research, scientific discovery, biology, materials science, neo-labs, AI models, drug discovery, scientific reasoning, Behnam Neyshabur, Harsh Mehta, Anthropic alumni, Andreessen Horowitz, Kleiner Perkins, AI simulation, molecular interaction, chemical space, computational science