About
We rebuilt LangGraph in Rust.
Then we kept the API. Then we open-sourced everything. That's the project.
Why we built it
We loved LangGraph's abstraction. State graphs are the right primitive for AI agents — explicit, checkpointable, debuggable, deterministic enough to test. What we didn't love was watching Python event-loops queue under load, paying for retry-and-recovery via SaaS tiers, and waiting for the Python runtime to import three megabytes of LangChain on cold start.
So we rewrote the runtime in Rust, kept the API surface intact, added the production primitives (circuit breakers, DLQs, distributed clusters) that everyone needs but no framework ships, and made it MIT so it stays free.
Who builds it
Agentmatic is a project by Neul Labs, a small team working on developer infrastructure. Maintainer: Dipankar Sarkar. We work in public on GitHub.
How we sustain it
The framework is, and will remain, open-source under MIT. We fund development by selling paid support tiers to teams who want SLAs — see /pricing. We will never relicense, never split into "community/enterprise editions," and never lock features behind a SaaS tier. If we go away, the code is yours forever.
Get in touch
- GitHub: github.com/neul-labs/agentmatic
- Email: agentmatic@neullabs.com
- LinkedIn: Dipankar Sarkar
Ship your next agent in minutes, not weeks.
MIT licensed. Drop-in for LangGraph. Native SDKs in 5 languages. Battle-tested resilience primitives in the box.