Langchain Basics

Premraj_MS

What is LangChain?

LangChain is an application framework that allows you to build applications using Large Language Models (LLMs). It acts as a bridge between LLMs and real-world applications.

Imagine this scenario: You want to plan a vacation to Dubai but are unsure about the places to visit within your budget. You ask ChatGPT for suggestions, and it provides you with an itinerary. However, you also want to automate tasks like setting travel dates, booking flight tickets, and reserving hotels. This is where AI agents come in, allowing you to perform these tasks seamlessly.

To build such an application, you can use LangChain.

Why Use Frameworks Like LangChain?

Building an AI-powered application from scratch requires extensive effort, including integrating APIs, handling data retrieval, and managing various AI models. Frameworks like LangChain simplify this process by providing predefined tools, APIs, and plugins that facilitate interactions with services like flight booking, restaurant reservations, web scraping, Gmail, and more.

LangChain does not make an LLM smarter; instead, it enables the LLM to interact with external systems and perform real-world actions efficiently.

While it is possible to build these functionalities manually from scratch, frameworks like LangChain save time, improve maintainability, and allow developers to focus on business logic rather than low-level implementations.

Core Components of LangChain

Chat Models: Supports LLMs like ChatGPT, DeepSeek, etc., enabling conversational AI capabilities.
Prompt Templates: Predefined structures for prompts, e.g.,
- "Write a summary about {topic} in a {tone}." This helps in dynamically generating responses based on user input.
Chains (Important): Workflows that break down complex tasks into smaller steps. Once a task is completed, the next step in the chain is triggered, ensuring a structured execution flow.
RAG (Retrieval-Augmented Generation) (Important): Enhances LLMs by retrieving and incorporating external data from PDFs, the internet, maps, GPS, etc., providing more contextually relevant responses.
Agents and Tools (Important):
- AI agents can autonomously make decisions and take actions based on user inputs.
- Tools allow integration with APIs and services, enabling real-world interactions like booking tickets, retrieving emails, or gathering data from websites.

By leveraging LangChain, developers can create intelligent, automated, and interactive applications with minimal effort compared to building such functionalities from scratch.

Premraj_MS

Pre-requisites for langchain

Python 3.8 or higher installed
Pycharm, VS Code, Sublime, etc, are preferred.
API access of any LLM (You can also use open-source AI models installed on your PC.)

Premraj_MS

First Leap to Langchain

Step 1: Create a folder and create a venv in it (Mac)

python3 -m venv <your venv name>

This will allow you to save and download files in a separate environment from a device. So the packages that you are installing on the venv can only be accessed within the same environment.

Step 2: Activate the Venv

source ./<your venv name>/bin/activate

Premraj_MS

Chat models in Langchain

A chat model in langchain is a component designed to communicate in a structured way with LLMs like GPT-4, Hugging Face, Claude Sonnet, etc…

Why Use LangChain Chat Models?

LangChain's chat models provide a structured way to interact with Large Language Models (LLMs), making it easier to build AI-powered applications. Here’s why they are useful:

1. Abstraction and Ease of Use

LangChain provides a simple interface to work with different LLMs (like ChatGPT, DeepSeek, Llama, etc.), so you don’t need to deal with low-level API calls or complex configurations.

2. Seamless Integration with Other Components

Chat models in LangChain can be easily combined with:

Prompt Templates (to standardize user queries)
Chains (to automate multi-step interactions)
Agents & Tools (to enable AI-driven actions)
Memory (to maintain conversation history)

3. Customization & Fine-Tuning

You can define response structures, fine-tune behavior, and integrate additional data sources (via Retrieval-Augmented Generation, RAG) to improve the model’s performance.

4. Handling Multi-Turn Conversations

Unlike a single API call, LangChain chat models maintain context across conversations, allowing AI to respond more naturally in ongoing discussions.

5. Supports Multiple Backends

LangChain lets you switch between different LLM providers (like OpenAI, Cohere, Hugging Face, etc.), giving flexibility based on cost, performance, and requirements.

6. Optimized for AI Agents

Chat models work well with AI agents, enabling dynamic decision-making, API calling, and real-time problem-solving based on user inputs.

In short, LangChain chat models simplify development, enhance flexibility, and improve AI capabilities, making them ideal for real-world applications. 🚀