Praxist is a new autonomous R&D tool that gives multiple AI coding agents a shared experiment loop: they can test ideas, evaluate results, keep track of failures, and use earlier findings to decide what to try next.
I tested it on the Adult Income dataset rather than a toy demo. It found a small repeatable feature-engineering improvement, rejected several worse options, and left behind a record of the experiments.
Full write-up:
https://medium.com/@rohanprichard/i-tried-praxist-by-sapient-labs-on-an-ml-dataset-ce6baecfa6ae
Top comments (0)