Micro1 is growing around an unexpected consequence of better AI: as models become more powerful, they need more sophisticated humans to teach them what good judgment actually looks like. Doctors, lawyers, engineers, finance professionals, and other specialists are increasingly being brought into model training and evaluation because the next generation of AI has to reason through complex real-world situations rather than simply predict the next word.
That demand has pushed Micro1’s reported gross annual run rate from $100 million to $500 million in eight months. Current reporting estimates the company retains roughly 60% to 70% of billed revenue, placing its net annual run rate closer to $150 million to $200 million. The figures are reported rather than independently confirmed.
Micro1 was founded in 2022 by Ali Ansari, who initially built the company around AI-powered recruiting. The business shifted after customers began using its talent platform to identify engineers and specialists for AI annotation work. That recruiting capability eventually became an advantage in a much larger market: finding, vetting, and managing experts who can create high-quality training data.
The company now positions itself as a data lab for frontier AI. Realm combines reinforcement-learning environments with expert-generated data and evaluations, while Cortex focuses on understanding how AI agents perform in production. Micro1 is also building robotics datasets designed to capture real-world human behavior and physical interactions.
Its infrastructure includes an AI recruitment engine that the company says has vetted more than 130,000 candidates across over 100 domains and 60 languages.
The growth suggests that AI training is moving beyond large volumes of generic labels. The scarcer resource is increasingly expert judgment: the ability to explain why one legal argument is stronger, why one medical interpretation is safer, or why an agent failed inside a realistic workflow.