Markov YC 지원서

Markov

Expert computer-use data for frontier AI labs

We collect and sell expert computer-use datasets designed for training and reinforcement learning tasks at frontier AI labs. Our product is a mix of proprietary and open-source data that helps labs train agents to use computers as well as humans can. Users download data from us, including 33,000+ hours we've directly supplied and open-source datasets that have been downloaded over 200,000 times on Hugging Face. This lets AI labs focus on models, not on wrangling large volumes of high-quality usage data.

San Francisco, CA, USA. We'll stay based here after YC.

We're in San Francisco to be close to our customers and the center of AI development. Our current and future customers are mostly here, so it's practical to stay.