Migration guide

From LangGraph to Agentmatic in 5 minutes.

One import change. Your StateGraph, your checkpointer, your tools — they keep working. You ship to production on a Rust engine that's 10× faster.

Step 1 — Install

pip install agentmatic

Step 2 — Change one import

# Before
from langgraph.graph import StateGraph, START, END
from langgraph.checkpoint.memory import MemorySaver

# After
from agentmatic import StateGraph, START, END
from agentmatic.checkpoint import MemorySaver

Every other line in your file stays the same. Re-run your existing tests; they should pass.

Step 3 — Bridge LangChain tools (if you use them)

from agentmatic.integrations.langchain import from_langchain_tools
from langchain_community.tools import WikipediaQueryRun, DuckDuckGoSearchRun

tools = from_langchain_tools([WikipediaQueryRun(), DuckDuckGoSearchRun()])
agent = create_react_agent(llm=llm, tools=tools)

Step 4 — Opt into resilience primitives

This is where you start getting value LangGraph can't give you without their paid platform tier:

agent = (Agent.builder("production")
    .llm(OpenAI("gpt-4o"))
    .tools([search, calculator])
    .checkpoint(S3Checkpointer(bucket="my-agents"))
    .circuit_breaker("api", failure_threshold=5)
    .retry_policy(RetryPolicy.exponential(max_attempts=3))
    .dead_letter_queue("failed")
    .interrupt_before(["dangerous_tool"])
    .build())

Step 5 — Optional: wrap as a LangChain Runnable

If you want to keep LangSmith tracing or use LangServe to deploy:

from agentmatic.integrations.langchain import as_langchain_runnable

runnable = as_langchain_runnable(agent)
# Now usable in LCEL, LangServe, LangSmith tracing — anywhere a Runnable goes.

Common gotchas

• Annotated reducers — Same syntax. Annotated[list, operator.add] works identically.
• Streaming — agent.stream(...) and agent.astream(...) both work. We also added agent.astream_events(...) for v2 LangChain-style events.
• Checkpoint formats — Memory and SQLite checkpoints are wire-compatible. Postgres requires a one-line schema migration we ship in agentmatic.checkpoint.postgres.upgrade().
• LangSmith — Set LANGCHAIN_TRACING_V2=true as usual; or use as_langchain_runnable() to keep traces.