Learning PathsLangChain AI Agent 从入门到精通

LangChain AI Agent 从入门到精通

通过一系列动手项目，系统学习使用 LangChain 构建 AI Agent。从基础的 LLM 调用开始，逐步掌握工具使用、记忆管理、RAG 检索增强，最终构建能够自主完成复杂任务的多 Agent 系统。

6 challenges

Beginner2 challenges

Hello, LangChain! Build Your First Conversational Chatbot

Using LangChain's ChatModel and PromptTemplate, build a command-line chatbot that supports multi-turn conversations from scratch. You'll learn how to combine an LLM, prompt templates, and conversation memory into a complete Chain, gaining a solid understanding of LangChain's core architecture.

2-3 小时

#01

Equip Your Agent with a "Toolbox"

Integrate multiple external tools — such as a search engine, calculator, and weather lookup — into a LangChain Agent, enabling it to automatically select and invoke the right tool based on user questions. Through this challenge, you'll master how to define, register, and use LangChain tools, ultimately building an intelligent Agent with true "action capabilities".

2-3 小时

Intermediate2 challenges

#02

Personal Assistant with Memory

Build a personal assistant with long-term memory using LangChain's Memory module. The assistant should be able to remember the user's name, preferences, and conversation history, and proactively leverage these memories in future interactions to deliver a personalized experience.

3-4 小时

#03

Local Knowledge Base Q&A System (RAG)

Build an intelligent Q&A system that supports PDF/text document uploads using LangChain's Document Loaders, text splitters, vector databases, and retrieval chains. Once a user uploads a document, the system can accurately answer questions based on the document's content rather than relying on the model's own training knowledge.

3-4 小时

Advanced1 challenges

#04

Autonomous Task Planning Agent

Using LangChain's ReAct (Reasoning + Acting) Agent framework, build an intelligent Agent that can receive high-level goals, autonomously decompose them into subtasks, formulate execution plans, and invoke multiple tools to progressively complete tasks. You will need to equip the Agent with tools such as search, calculation, file read/write, and implement full chain-of-thought tracking and task state management.

4-6 小时

Expert1 challenges

#05

Multi-Agent Collaboration System

Design and implement a multi-agent pipeline consisting of a 'Researcher', 'Writer', and 'Reviewer' that work together to complete the task of generating a high-quality industry report.