AI / LangGraph LangChain Interview questions
How does state management work in LangGraph?
State in LangGraph is a TypedDict that is shared across all nodes in a graph run. Every time a node executes, it can return a partial update — a dict containing only the keys it wants to change. LangGraph merges the update into the current state using reducers.
The default reducer is last-write-wins: the node's returned value replaces the current value for that key. You can override this with Annotated[type, reducer_fn] where reducer_fn takes (current, update) and returns the new value:
from typing import TypedDict, Annotated import operator class GraphState(TypedDict): # Append-only: new messages are added to the list messages: Annotated[list, operator.add] # Last-write-wins: iteration count is replaced each time iteration_count: int # Custom reducer: keep the highest score seen so far best_score: Annotated[float, lambda a, b: max(a, b)]
State is immutable between node calls — nodes receive a snapshot and return updates; they do not mutate state in place. This design enables checkpointing (save the full state after each node), time-travel debugging (replay from any past state), and parallel node execution (each branch gets a copy of the state).
More Related questions...
