LangChain.js Guide

LangChain is a framework for developing applications powered by language models. It excels at orchestrating complex workflows, managing memory, and connecting to external data.

Core Concepts (LCEL)

Modern LangChain uses LCEL (LangChain Expression Language) to compose chains using pipe | syntax.

import { ChatOpenAI } from "@langchain/openai";
import { PromptTemplate } from "@langchain/core/prompts";
import { StringOutputParser } from "@langchain/core/output_parsers";

const model = new ChatOpenAI({ model: "gpt-4o" });
const prompt = PromptTemplate.fromTemplate("Tell me a joke about {topic}");
const outputParser = new StringOutputParser();

// The Chain
const chain = prompt.pipe(model).pipe(outputParser);

// Run it
const response = await chain.invoke({ topic: "ice cream" });
console.log(response);

Chains vs. Agents

• Chains: Hardcoded sequences of steps. (Input -> Prompt -> LLM -> Output)
• Agents: The LLM decides what steps to take. (Input -> LLM -> Tool A -> LLM -> Tool B -> Output)

Building an Agent

Agents can use "Tools" to interact with the world.

import { ChatOpenAI } from "@langchain/openai";
import { createOpenAIFunctionsAgent, AgentExecutor } from "langchain/agents";
import { Calculator } from "langchain/tools/calculator";

const model = new ChatOpenAI({ temperature: 0 });
const tools = [new Calculator()];

const agent = await createOpenAIFunctionsAgent({
  llm: model,
  tools,
  prompt: /* ... standard agent prompt ... */,
});

const executor = new AgentExecutor({
  agent,
  tools,
});

const result = await executor.invoke({
  input: "What is 5 to the power of 3?",
});

Memory

LLMs are stateless. LangChain provides memory to store chat history.

Note: In production, it's often better to manage history in your own database and pass it as context, rather than relying on in-memory storage.

Integration with Next.js

You can use LangChain to generate the response, and then stream it using the Vercel AI SDK.

// app/api/chat/route.ts
import { ChatOpenAI } from '@langchain/openai';
import { LangChainAdapter } from 'ai';

export async function POST(req: Request) {
  const { messages } = await req.json();

  const model = new ChatOpenAI({ model: 'gpt-4o' });

  const stream = await model.stream(messages);

  return LangChainAdapter.toDataStreamResponse(stream);
}

When to use LangChain?

• ✅ Complex multi-step chains (e.g. Summarize text -> Translate -> Extract keywords).
• ✅ Switching between models (easy to swap OpenAI for Anthropic).
• ✅ Using pre-built tools (Calculators, Web Search, SQL).

Next Steps

• Explore LlamaIndex for data-heavy applications.