Getting Started with LangChain: A Fun Guide for Absolute Beginners

Author(s): Kapil Deshmukh

Originally published on Towards AI.

Imagine you’re building with LLMs (large language models), but every time you need prompts, memory, or tool integration, you reinvent the wheel. Enter LangChain — your LLM-powered Swiss Army knife.

Large Language Models (LLMs) like OpenAI’s GPT, DeepSeek, or Anthropic’s Claude are powerful — but let’s be honest, using them directly can get messy. Imagine writing raw prompts for every task, managing APIs, memory, vector databases, and agents on your own. 😅 That’s where LangChain comes in.

LangChain is like the Lego toolkit for LLMs: it gives you ready-made building blocks so you can quickly create AI apps without reinventing the wheel.

In this article, we’ll explore:
✅ What LangChain is
✅ Why developers use it
✅ Key components explained with examples
✅ A mini hands-on project

🌟 Why LangChain?

Think of an AI app you might want to build:

• A chatbot that remembers past conversations
• A research assistant that searches PDFs or the web
• An agent that can call APIs or execute code

Without LangChain, you’d spend weeks wiring everything manually. With LangChain, you snap components together like blocks.

🧩 Key Concepts in LangChain (Explained Simply)

Here are the “Lego blocks” you’ll use most often:

• LLMs → The brain (e.g., GPT, DeepSeek, LLaMA).

from langchain_openai import OpenAI

llm = OpenAI(temperature=0.7, openai_api_key="your_key")

This sets up your AI brain with a creative flair (temperature = randomness).

• Prompts → The instructions you give to the model.

from langchain.prompts import PromptTemplate

prompt = PromptTemplate.from_template("Write a tweet about {topic}")

Prompts are like recipes. You tell the AI what ingredients (variables) to use.

• Chains → Linking prompts + LLMs into a pipeline.

from langchain.chains import LLMChain

chain = LLMChain(llm=llm, prompt=prompt)
print(chain.run("LangChain basics"))

Chains let you connect multiple steps, like asking → answering → summarizing.

• Memory → Letting the chatbot “remember” past chats.

from langchain.memory import ConversationBufferMemory

memory = ConversationBufferMemory()

Without memory, your AI is like Dory from Finding Nemo 🐟

• Agents → Giving AI tools (search, calculator, APIs).

# Pseudocode
agent = initialize_agent(tools=[search_tool, calculator], llm=llm)
agent.run("What's 2+2 and latest AI news?")

• Vectorstores → Long-term memory for documents.

from langchain_community.vectorstores import FAISS

vectorstore = FAISS.from_texts(docs, embeddings)

Useful for RAG (Retrieval-Augmented Generation): answering from your PDFs, websites, or databases.

🚀 A Hands-On Mini Project

from langchain_openai import OpenAI
from langchain.chains import RetrievalQA
from langchain_community.vectorstores import FAISS
from langchain.embeddings import OpenAIEmbeddings

# Step 1: LLM
llm = OpenAI(openai_api_key="your_key")

# Step 2: Create embeddings
embeddings = OpenAIEmbeddings(openai_api_key="your_key")

# Step 3: Store notes in vector DB
docs = ["LangChain is a framework for LLMs.",
 "It helps build chatbots, agents, and RAG systems."]
db = FAISS.from_texts(docs, embeddings)

# Step 4: Retrieval + QA
qa = RetrievalQA.from_chain_type(llm=llm, retriever=db.as_retriever())
print(qa.run("What is LangChain?"))

This is RAG in action, one of the most practical use case today.

👉 If you found this useful, clap 👏 and follow me for more AI tutorials!

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI