What Thesis does

Thesis is an applied AI research lab building systems that automate the process of AI research and development itself. Its engine handles parts of the machine-learning workflow, such as planning experiments, training and evaluating models, and iterating on designs, with the goal of making AI discovery faster and cheaper.

Key capabilities

  • Automates planning and execution of ML experiments
  • Trains and evaluates models autonomously
  • Iterates on model designs as part of an R&D engine

Who it's for

Thesis is aimed at AI research and machine-learning teams seeking to accelerate model development and discovery. Its differentiation in the AI search and research space is automating the R&D loop end to end; the company reported state-of-the-art results on OpenAI's MLE-Bench, which tests how well AI systems can train ML models autonomously. Thesis was founded by a former Google X, Nvidia, and Stanford AI Lab researcher and is part of Y Combinator's Fall 2025 batch.