Biomni

Why we included this project

Biomni suits scientists and ML engineers who would rather spend time on experiments than on wiring together the sequential steps of a research task. It combines LLM reasoning with retrieval and code-based execution, so it can take a plain-language request like 'design a CRISPR screen for T cell exhaustion', run the steps, pull in relevant data, and return a testable result. Tool selection, data fetching from its bundled datalake, and analysis pipelines all live inside the agent, so you do not have to assemble them yourself. It ships a Python API and a Gradio web UI, letting you run batch work from a notebook or hand the interactive interface to a lab for ad-hoc questions. For teams exploring autonomous scientific workflows, it is an installable starting point with a config system for swapping model providers.