← Back to Projects

Project · mit-catalyze

Catalyze

1st Place - MIT Site LLM Hackathon for Chemistry & Materials2 stars

AI-powered chemistry assistant with a multi-agent architecture. Specialized agents handle research, protocol generation, lab automation (OpenTrons & Lynx), and safety analysis, backed by ChEMBL database integration.

PythonFlaskGPT-4oChEMBLMulti-Agent
mit-catalyze
Catalyze

About

Catalyze uses a multi-agent architecture where specialized agents handle distinct aspects of chemistry workflows. The Research Agent searches ChEMBL databases and literature, the Protocol Agent generates detailed experimental procedures, the Lab Automation Agent interfaces with OpenTrons and Lynx instruments, and the Safety Agent performs hazard analysis on proposed experiments. Built in 24 hours at MIT.