300 GitHub stars and 6,000 crates.io downloads later, graph-flow has been a fun experiment in building AI-agent workflows in Rust.

It is a LangGraph-inspired library focused on graph-based orchestration, stateful execution, conditional routing, persistence, and type safety. The goal was to keep the core small and useful not only for AI agents, but for any Rust application that needs robust workflow execution.

Rust in AI still feels early, but projects like Rig, LanceDB, and others make it feel like the ecosystem is starting to take shape.

If you’re building AI systems in Rust, I’d love for you to check it out and share feedback.

GitHub: https://github.com/a-agmon/rs-graph-llm