Mirendil Raises $200M to Build Self-Improving AI Platform

Editorial analysis: For AI practitioners, the race to automate parts of the ML research loop is both a technical lever and a governance inflection point: wider access to automated experiment design and model iteration could accelerate specialised research but also concentrates powerful capabilities outside established lab safeguards. Reported facts: According to The Next Web and TechFundingNews, Mirendil, founded by former Anthropic researchers Behnam Neyshabur and Harsh Mehta, raised $200 million in a seed round at a $1 billion valuation co-led by Andreessen Horowitz and Kleiner Perkins with participation from NVIDIA. Gizmodo reports the San Francisco startup has about 20 technical staff, job postings offering up to $500,000, and a stated aim to build systems that automate AI research, a concept often called recursive self-improvement. WSJ and other coverage frame the company as targeting labs, universities, and organisations that lack in-house ML R&D teams.