Learn retrieval augmented generation, brick by brick.
14 lessons
Automatically mirror LangChain's YouTube Channel. Not run by the LangChain team. The official YouTube channel is https://www.youtube.com/@LangChain and Academy https://academy.langchain.com/
This lesson teaches you to build interactive chatbots using LangChain's LangGraph and React. It covers setting up a LangGraph project, integrating the `useStream` hook for real-time conversation management, and building a user interface with features like message editing and branch switching for enhanced user experience.
This lesson explores two key approaches to AI-powered deep research: flexible, agent-based methods and structured, workflow-based systems. It highlights the trade-offs between these approaches, showcasing LangChain's open-source tools and their role in building efficient and cost-effective deep research workflows.
This lesson explores how AI agents utilize different memory types—semantic, episodic, and procedural—similar to human memory, leveraging LangMem and LangGraph libraries within the Langchain framework. By efficiently storing and retrieving this information, agents personalize interactions, learn from past experiences, and adapt their behavior for improved performance.
This lesson demonstrates building intelligent agents with robust memory management using LangChain and LangSmith. By leveraging tools like `manage_memory` and `search_memory`, agents learn and adapt from past interactions, maintaining context across conversations and improving efficiency.
This lesson teaches you to build a sophisticated email assistant agent using LangChain's multi-agent system and prompt optimization. The agent learns from user feedback, improving its email drafting skills (including adding signatures and meeting options) and task routing efficiency through a supervisor agent.
LangGraph Agents and Workflows
7 lessons
Master LangChain's agentic applications by learning how memory and state management enhance chatbot performance. This module explores efficient message history handling, database integration (like Postgres and SQLite), and techniques to optimize context windows.
This lesson teaches you to build memory-enabled agents in LangGraph, deploying them via LangGraph Cloud. It then shows how to manage agent state effectively using Python's `TypedDict`, `dataclass`, and Pydantic for flexible schema definition, type hinting, and runtime validation.
This lesson explores LangGraph's state management, showcasing how to define and update state schemas using Python's `TypedDict` and handle concurrent state updates using reducers and the `Annotated` type. It further demonstrates efficient message management with the `add_messages` reducer, including adding, updating, and removing messages using unique IDs.
This lesson teaches how to manage complex data flows in LangGraph using multiple schemas within graph nodes. It demonstrates using `OverallState` and `PrivateState` TypedDicts to control internal node communication and filter outputs, enhancing graph flexibility and code clarity.
This lesson teaches you to build efficient chatbots with long-term memory using Langchain and LangGraph, focusing on managing message history to avoid exceeding token limits. Key techniques include customizing state schemas, employing reducers like `RemoveMessage` and `add_messages`, and leveraging LangSmith for debugging and tracing.
This lesson teaches you how to build a conversational AI chatbot with memory using Langchain and LLMs. It covers techniques for summarizing conversations, managing message history efficiently, and leveraging checkpoints for long-running interactions, all while demonstrating practical applications with real-world examples.
Master seamless human-agent collaboration by strategically integrating human oversight into automated workflows using breakpoints. LangChain Academy's Module 3 teaches you how to manage sensitive tasks, delegate work, and debug effectively through human-in-the-loop techniques.
This lesson teaches you how to build interactive chatbots using LangGraph's streaming capabilities. It covers different streaming modes ("updates" and "values"), demonstrates handling streamed responses from LLMs like GPT-4, and shows how to access and process individual tokens in real-time.
