📄️ Prompting Agents: System Prompts and Structured Output
Learn how to shape an AI agent's behavior with system prompts, few-shot examples, and guaranteed structured output, building a real email assistant from scratch.
📄️ Tool Calling: Giving Agents Abilities
Learn how to give an AI agent real abilities using LangChain's @tool decorator, and understand exactly how an agent decides when to use one.
📄️ Web Search: Real-Time Knowledge for Agents
Learn why language models can't know current events on their own, and how to give an AI agent real-time web search using the same tool-calling pattern you already know.
📄️ Memory and Threads: Agents That Remember
Learn why AI agents forget everything between calls by default, and how to give them real conversation memory using LangGraph checkpointers and threads.
📄️ Multimodal Messages: Text, Image, and Audio Input
Learn how to send images and audio to an AI agent alongside text, and understand the base64 encoding that makes it possible.
📄️ MCP: Connecting Agents to External Servers
Learn what the Model Context Protocol (MCP) is and how to connect an AI agent to both your own custom MCP servers and ones built by other people.
📄️ Runtime Context: Injecting User-Specific Data
Learn how to give an AI agent fixed, user-specific information at invocation time using LangChain's context_schema, and how it differs from memory.
📄️ Custom Agent State: Reading and Writing Beyond Messages
Learn how to extend an AI agent's state with your own custom fields that tools can read and write during a conversation, and how this relates to memory.
📄️ SQL Agents: Querying Databases in Natural Language
Learn how to let an AI agent write and run its own SQL queries against a real database, safely and with proper error handling.
📄️ RAG Agents: Retrieval-Augmented Answers
Learn how to let an AI agent search an unstructured document like a PDF handbook, using embeddings and semantic search, the foundation of retrieval-augmented generation (RAG).
📄️ Multi-Agent Systems: Subagents and Delegation
Learn two patterns for splitting work across multiple AI agents: wrapping subagents as tools, and a coordinator that delegates to specialists through shared state.
📄️ Managing Conversation History: Summarization and Trimming
Learn what middleware means in LangChain agents, and how to keep long conversations efficient using automatic summarization and message trimming.
📄️ Human-in-the-Loop: Approve, Reject, Edit
Learn how to pause an AI agent before it takes an irreversible action, and let a human approve, reject, or edit it using LangChain's HumanInTheLoopMiddleware.
📄️ Dynamic Middleware: Adapting Prompts, Tools, and Models at Runtime
Learn how to make an AI agent's prompt, available tools, and underlying model all adapt automatically based on context and conversation state.
📄️ Capstone: A Production-Ready Authenticated Email Assistant
Assemble everything from this series into one complete, production-ready version of Aria — authenticated, context-aware, dynamically adaptive, and human-supervised.