Agent framework

OpenBook vs LangGraph

LangGraph is a powerful code library for stateful agent graphs. OpenBook is a self-hostable platform that gives you the same graph power — branching, state, human-in-the-loop — through a UI and an AI builder, plus persistence, a plugin ecosystem, and core memory. LangGraph is for engineers building in Python/JS; OpenBook is for teams who want to operate workflows without owning the orchestration code.

At a glance

OpenBook and LangGraph, capability by capability.

CapabilityOpenBookLangGraph
Open source
Visual builder + AI builder
Stateful graph execution
Conditional branching
Human-in-the-loopInterrupts (code)
Runnable product (UI, persistence)
Embed as a code library

Why teams pick OpenBook over LangGraph

No orchestration code

Branching, variables, error paths, and human pauses are nodes on a canvas — or generated by the AI builder — not graph code you write and maintain.

A product around the graph

UI, run history, an approvals inbox, settings, and a plugin system ship in the box.

Core memory + actions pool

Agents remember by default and draw from a shared pool of actions; activating a plugin extends it.

OpenBook

MIT platform, self-host.

LangGraph

MIT library; LangGraph Platform is a paid hosted offering.

When to choose which

Choose OpenBook when…
  • You want the graph power without writing and maintaining graph code
  • You want a UI, persistence, and approvals out of the box
Choose LangGraph when…
  • You're an engineering team building a bespoke agent app in code
  • You need fine-grained, code-level control of every state transition
Switching from LangGraph

Translate LangGraph nodes/edges to OpenBook agent and action nodes; interrupts become human.handoff nodes; channels/triggers become trigger nodes.

Frequently asked questions

Build your first agentic workflow today.

OpenBook is free and open source. Self-host it in minutes, or read the docs to see how far the engine goes.