📰 Autonomous AI News Agent

An end-to-end AI-powered system that autonomously searches, summarizes, and publishes news articles using Mistral-7B and the Google Blogger API.

📊 Project Showcase

🎬 Watch the AI News Agent in Action:

WhatsApp.Video.2025-02-24.at.19.01.49.mp4

📸 Example AI-Generated Blog Post:

🔄 End-to-End Pipeline:

---

🚀 Overview

Our AI-driven solution automates the entire news publishing process by integrating real-time web scraping, NLP-based content processing, and automated publishing. It collects news from various sources, including Times of India, processes the content using Mistral-7B for structured summaries, and enhances SEO with SpaCy and YAKE. A SQLite database ensures duplicate filtering, while the final output is formatted in JSON for structured storage. The system then publishes articles automatically to a blog via the Google Blogger API, creating a fully autonomous pipeline that transforms raw web data into polished, SEO-optimized articles in minutes.

---

🔧 How It Works

Step 1: Web Scraping for News Collection 🕵️‍♂️

• Fetches news articles from sources like Times of India using Tavily Web Search API.
• Uses BeautifulSoup for HTML parsing and data extraction.

---

Step 2: AI-Based Summarization & SEO Enhancement 🤖

• Mistral-7B (4-bit quantized) generates structured summaries.
• SpaCy & YAKE optimize the content for SEO.
• Filters duplicate content using SQLite.

---

Step 3: JSON Formatting for Structured Storage 🗄️

• The final output is stored in JSON format.
• Ensures compatibility with the Blogger API.

📌 Example JSON Output:

{
  "title": "Latest Mumbai Crime: Bank Robbery in Bandra and Police Investigation",
  "summary": "A bank in Bandra was robbed early this morning. The police have launched an investigation...",
  "published_date": "2025-02-24"
}

---

Step 4: Automated Publishing to Blogger ✍️

• Uses Google Blogger API to auto-publish posts.
• Formats posts with rich text, images, and structured data.

---

🔑 Required Credentials

Credential	Purpose
Hugging Face Token	Access Mistral-7B for AI-based summaries
Tavily API Key	Web search and scraping
Google Credentials	Blogger API authentication

🔹 Setup Your Credentials

1. Hugging Face Token (For Mistral-7B)

   • Create an account at huggingface.co
   • Get a token from Settings → Access Tokens
   • Add to the code: login(token="YOUR_HF_TOKEN")

2. Tavily API Key

   • Sign up at tavily.com
   • Add to the code: tavily_api = "YOUR_API_KEY"

3. Google Blogger API Credentials

   • Create a Google Cloud Project
   • Enable Blogger API
   • Create an OAuth 2.0 Client ID
   • Download credentials as configure.json

---

📥 Installation

git clone https://github.com/yourusername/daksh-mor-blog-agent.git
cd daksh-mor-blog-agent
pip install -r requirements.txt

---

🛠️ Usage

1. Configure Credentials in pipeline.ipynb
2. Set Target Topic in the notebook
3. Run All Cells to:
   • Scrape latest news
   • Generate SEO-optimized summary
   • Publish to Blogger automatically

---

⚙️ Tech Stack

Component	Technology Used
AI Model	Mistral-7B (4-bit quantized)
Web Scraping	Tavily Web Search API, BeautifulSoup
SEO Optimization	SpaCy, YAKE
Storage	SQLite (for URL tracking)
Publishing	Google Blogger API

---

🌟 Features

✅ Fully automated news generation pipeline
✅ AI-powered summarization & SEO optimization
✅ Supports multiple news sources
✅ JSON-formatted structured storage
✅ Auto-publishing with Google Blogger API

---

🎯 Future Enhancements

🔹 Support for more news sources
🔹 Multi-language news summarization
🔹 Integration with other publishing platforms (e.g., WordPress, Medium)

---

📌 License

This project is licensed under the MIT License - see the LICENSE file for details. pi lovers though.