AI Browser Agent

This repository demonstrates the creation of simple and autonomous AI-powered web agents that can interact with websites, automate browsing tasks, extract structured data, and perform custom actions using Python and modern LLM (Large Language Model) APIs.

Features

• Web Scraping: Automates the extraction of structured data (such as course information) from web pages.
• Screenshots: Captures screenshots of web pages during automation.
• LLM Integration: Uses OpenAI and other APIs to guide browsing and data extraction tasks.
• Agent Architectures: Provides both simple and autonomous agent examples.
• Notebook-based: All code and demonstrations are provided in Jupyter notebooks for easy experimentation.

Notebooks

• Building a Simple Web Agent.ipynb: Step-by-step guide to building a basic web scraping and automation agent using Playwright and OpenAI.
• Building an Autonomous Web Agents.ipynb: More advanced, multi-step autonomous web agent example (including integration with the MultiOn API).

Installation

1. Clone the repository

   git clone https://github.com/injamul3798/ai_browser_agent.git
   cd ai_browser_agent

2. Install dependencies

   • Make sure you have Python 3.11+ installed.
   • Install the required packages:

     pip install -r requirements.txt

3. API Keys

   • Set your OpenAI API key as an environment variable or in the notebook as prompted.
   • For autonomous agents, you may need additional API keys (e.g., MultiOn).

Usage

1. Open the desired notebook (.ipynb) in Jupyter Notebook or JupyterLab.
2. Follow the step-by-step instructions within the notebook cells.
3. Modify target URLs and instructions as needed for your automation tasks.

File Guide

• requirements.txt: Python dependencies.
• helper.py: Utility functions for API keys, displaying results, etc.

Example Use Case

• Scrape and summarize online course listings.
• Take automated screenshots of web pages.
• Interact with and automate multi-step browser tasks using LLMs.

Resources

• Building AI Browser Agents by DeepLearning.AI

Tips

• Download notebooks via File > Download as > Notebook (.ipynb) in Jupyter.
• See included markdown cells for troubleshooting and additional guidance.

License

This project is for educational purposes and currently does not specify a license. Please add one if you plan to use or distribute this code.

---

Author: injamul3798