Building AI Applications with LangChain and Node.js

Go beyond simple AI chains. Learn to orchestrate complex, multi-agent systems and debug them effectively using the modern Node.js stack.

#1about 4 minutes

Defining modern AI applications and core concepts

An overview of what constitutes a generative AI application and a review of fundamental concepts like LLMs, inference, context windows, and model evaluation.

#2about 2 minutes

Exploring common AI application patterns

A breakdown of the four primary architectural patterns for AI applications: chat, retrieval-augmented generation (RAG), single-agent, and multi-agent systems.

#3about 2 minutes

Understanding the modern LLM application stack

A look at the key components of the LLM stack, including the agent runtime, inter-agent communication, data retrieval with vector databases, and LLMops.

#4about 1 minute

Introducing key protocols for agent communication

An explanation of the Model Context Protocol (MCP) for extending agent context and the Agent-to-Agent (A2A) protocol for enabling communication between different agents.

#5about 3 minutes

How to choose the right tools for your AI application

Guidance on selecting the appropriate tools for your project, including programming language, LLM provider, and vector database, with a focus on Node.js and PostgreSQL.

#6about 2 minutes

Getting started with LangChain for Node.js

An introduction to the LangChain.js ecosystem, covering its core packages, community integrations, and the powerful LangChain Expression Language (LCEL) for composing chains.

#7about 2 minutes

Building complex agents with LangGraph

Learn when to use LangGraph instead of standard LangChain for building complex, stateful multi-agent systems with branching logic and retry mechanisms.

#8about 5 minutes

Composing a basic chain with the expression language

A practical example of how to use the LangChain Expression Language (LCEL) to pipe together a prompt template, an LLM, and an output parser in a few lines of code.

#9about 4 minutes

Live code demo of various AI application patterns

A walkthrough of runnable code examples demonstrating structured output, chat with memory, retrieval-augmented generation (RAG), and a multi-agent supervisor architecture.

#10about 1 minute

Using LangSmith for observability and debugging

An overview of how LangSmith provides essential observability, allowing you to trace, debug, and evaluate the performance of complex agent and chain executions.