MoDesign

MolChat is an open-source, extensible AI laboratory assistant for early-stage in silico drug discovery, designed to transform natural-language instructions into real, verifiable computational workflows. Computational drug discovery often requires researchers to manually coordinate multiple tools for molecular preparation, docking, structural analysis, quality control, visualization, and documentation. These fragmented workflows can be time-consuming, technically demanding, and difficult to reproduce. MolChat addresses this by allowing researchers to describe their scientific objectives in natural language. The agent can plan multi-compound docking experiments, request human approval, execute scientific tools, analyze computational results, compare molecular poses, generate molecular animations, and export reproducible evidence packages. MolChat is not tied to MolDesign. It is designed as a modular, platform-independent agent architecture that can be extended through custom tools and integrations. MolDesign serves as its first real-world implementation and scientific use case, but developers can adapt MolChat to interact with their own applications, computational engines, and specialized workflows. What was built during the hackathon? MolDesign, my desktop platform for early-stage drug discovery, existed before the event. During the hackathon, I evolved MolChat from a basic conversational assistant into a tool-driven scientific agent capable of coordinating real operations across the platform, with human oversight and evidence-backed interpretation. The long-term vision is to make MolChat a reusable foundation for trustworthy scientific agents, rather than a chatbot restricted to a single application. The LLM orchestrates. Scientific engines calculate. Evidence is preserved.