Potato

Potato is a scientific operating system for life sciences, built agents first and powered by AI agents that design, analyze, and run experiments. Its research collaborator, called Tater, reasons across scientific content and data, reads the literature, forms hypotheses, designs experiments, generates protocols, writes and runs bioinformatics scripts, drives lab robots for automation, analyzes raw data, interprets results, and decides what to run next, closing the loop from idea to experimentally validated result. Potato runs on purpose built cloud infrastructure that autoscales compute and GPUs, so teams can branch into alternative research directions and evaluate hundreds of variations in parallel, and its tools can communicate directly with each other to extend the duration of tractable tasks. Grounded in peer reviewed literature through retrieval augmented generation, and focused on reproducibility, Potato automatically produces a summary report and an LLM as judge scoring rubric for completion, correctness, and efficiency after each agent session. Its first product, the Optimizer, handles closed loop optimization of plate based experiments.