LangGraph

An open-source agent-runtime framework built on top of LangChain that models agentic workflows as **state machines** — supervisor/subagent topologies, branching reasoning paths, error-recovery loops, and durable checkpoint persistence. LangGraph has become the industry-standard pattern for production agentic deployments where workflows must reliably pause, audit, and resume after failures or long-running operations. Built-in checkpointer abstractions integrate with multiple backends including S3-compatible object storage.

Naive agentic loops (LLM call → tool call → repeat) fail at production scale because they have no durable state, no error-recovery semantics, and no audit trail. LangGraph imposes structure: nodes are explicit, edges encode control flow, and the framework's checkpointer captures state at every transition. When an agent crashes mid-workflow, LangGraph can resume from the last checkpoint — which is the basic property that turns a flaky agent demo into operationally deployable infrastructure.

Multi-agent supervisor topologies, durable agentic workflows with mid-task pause/resume, error-recovery branching, audit-grade trace persistence, S3-backed checkpointer for cross-region agent durability, FaaS-decomposed agentic patterns (FAME-style Planner / Actor / Evaluator).