LangGraph
LangGraph is a framework for building stateful, graph-structured workflows and applications, including those involving language models and agents.
Build Python Web Apps with Streamlit
LangGraph is a framework for building stateful applications as workflows composed of interconnected graph nodes, including applications driven by language models.
- How it is used
- Structures a customer-service chatbot’s retrieval, conversation, and order actions as a workflow.
- How deep it goes
- Builds from graph fundamentals to a one-node chatbot, checkpointed memory, and conditional tool routing.
- Examples
- Uses a one-node chatbot, remembered conversation, and order-status tool calls.
- What you can do after
- Build it from scratch as a stateful chatbot graph, then apply ToolNode and tools_condition to route actions.
Read how Build Python Web Apps with Streamlit explains LangGraph →
AI Agents and Applications
LangGraph is a LangChain module for representing application workflows as graphs with branching and more complex control flow.
- How it is used
- Organizes stateful agentic workflows that branch, retry, and coordinate tasks beyond a linear LangChain chain.
- How deep it goes
- Works through typed state, node functions, conditional routing, compilation, and invocation, then contrasts manual graphs with prebuilt agents.
- Examples
- Uses a web research assistant that retries weak searches and a travel information agent that calls tools.
- What you can do after
- Build it from scratch with StateGraph, typed state, nodes, and conditional edges, or apply it via a prebuilt ReAct agent.
Hugging Face in Action
LangGraph is a framework for building stateful, multi-step applications whose components communicate through graph-based state transitions.
- How it is used
- Builds stateful, multistep LLM agents that reason, branch, use tools, call APIs, and preserve conversational context.
- How deep it goes
- Explains nodes, edges, ReAct execution, tools, message state, and memory, while focusing only on core agent-workflow components.
- Examples
- Demonstrates a Bill Gates question, arithmetic, election search, weather lookup, custom tools, and follow-up conversation.
- What you can do after
- Build agents with create_react_agent(), connect custom or external tools, inspect traces, and maintain conversations through explicit state.
Knowledge Graphs and LLMs in Action
LangGraph is a framework for building stateful, graph-structured workflows and agent applications powered by language models.
- How it is used
- Orchestrates stateful, multi-agent language-model workflows for knowledge-graph question answering and interactive visualization.
- How deep it goes
- Defines shared state, graph nodes, fixed and dynamic edges, then implements intent, schema, query, execution, retry, and summarization agents.
- Examples
- Demonstrates document retrieval, answer generation, knowledge-graph querying, visualization selection, retries, summaries, tables, and maps.
- What you can do after
- Build stateful QA pipelines with shared state, conditional routing, error retries, generated summaries, and interactive Streamlit output.
Read how Knowledge Graphs and LLMs in Action explains LangGraph →
Machine Learning Platform Engineering
LangGraph is a framework extension for orchestrating language-model applications as graphs with branching, loops, and stateful transitions.
- How it is used
- Orchestrates agentic language-model workflows when sequential chains cannot express runtime decisions, branching, or iterative refinement.
- How deep it goes
- Contrasts sequential and graph-based architectures, but stops short of implementation code or graph-construction details.
- Examples
- Demonstrates intent routing among cache, search, and clarification paths followed by response validation and refinement loops.
- What you can do after
- Apply LangGraph to design language-model workflows with branching, parallel paths, and iterative response improvement.
Read how Machine Learning Platform Engineering explains LangGraph →
The Art of AI Product Development
A framework for building stateful AI agents and graph-structured workflows with support for complex interactions.
- How it is used
- Builds sophisticated, stateful AI agents and graph-based workflows for complex interactions.
- How deep it goes
- Defines graph architecture and key capabilities but stops short of implementation detail.
- Examples
- Illustrates state management, multi-turn conversations, and LangChain integration as representative capabilities.
- What you can do after
- Build stateful AI agents with graph-based workflows, state management, and multi-turn conversations.
Read how The Art of AI Product Development explains LangGraph →
Building Reliable AI Systems
LangGraph is a library for building stateful, graph-based applications with language models and agents.
- How it is used
- Coordinates specialized agents in applications that combine capabilities such as search, policy retrieval, and vision.
- How deep it goes
- Defines graph primitives and capabilities, but does not provide implementation details or operational procedures.
- Examples
- Uses a hiking-boots photo request combining vision, product search, and return-policy retrieval.
- What you can do after
- Build it from scratch: design custom multi-agent workflows with explicit states, nodes, edges, and routing decisions.
Related concepts
- CrewAIalternative to · CrewAI is an open-source framework for building and orchestrating collaborative AI agents and multi-agent workflows.
- state managementuses · State management is the practice of organizing, maintaining, and updating information that represents a system’s or process’s condition over time.
- error handlinguses · Error handling is the practice of detecting, representing, propagating, reporting, and responding to errors or failures during software operation.
- multi-agent systemuses · A multi-agent system is a computational system composed of multiple agents that interact or coordinate to perform tasks or pursue goals.
- branchuses · Branching is the selection among alternative execution paths according to a condition, outcome, or program state.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.