Use case
Production LangGraph replacement
You already shipped on LangGraph. Keep the code, change one import, and run it on a Rust engine that's built for production load.
In one line: Production users report 8–12× p95 latency drops on multi-agent workloads, with 50–80% less memory in the benchmark suite. See the numbers on the benchmarks page.
The problem
- Your graphs work, but p95 latency and memory grow with every agent and tool you add.
- Resilience (retry, circuit breaking, distributed execution) lives behind LangGraph Platform's paid tier.
- A full rewrite onto another framework is a risk you can't justify.
How Agentmatic solves it
One import changes
Replace from langgraph.graph with from agentmatic. StateGraph, conditional edges, checkpointers, time travel and HITL keep the same API.
Resilience in the open-source core
Circuit breakers, retry policies and dead-letter queues ship in the box — features that require LangGraph Platform elsewhere.
Keep your ecosystem
from_langchain_tools() bridges your LangChain tools; as_langchain_runnable() keeps LangSmith tracing, LCEL and LangServe working.
Keep reading
Part of the Agentmatic use cases, from Neul Labs ↗.
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.