MattGPT

Agentic AI Research Platform

Welcome to MattGPT

Your intelligent research companion powered by IBM's Granite 3.2 LLM. Experience Two-Track Architecture combining chain operations for complex workflows with enhanced chat for conversational interactions, all running locally on your machine.

MattGPT: Local Agentic AI Research Platform

About MattGPT

Free and open-source research automation powered by local AI models

🎯 Two-Track Architecture

🔗 Chain Operations Track
Button-driven complex workflows including "What's Going On?" analysis, Week in Review generation, and Daily Podcast creation with progress tracking.
💬 Enhanced Chat Track
Conversational interactions with article context, customer database queries, URL analysis, and PDF processing capabilities.
🧠 Local AI Intelligence
Powered by IBM's Granite 3.2:8b model running locally via Ollama, providing summarization, relevance scoring, and content analysis.
🎙️ AI-Generated Content
Edge-TTS voice synthesis creates daily podcasts with dual speaker personalities and intelligent script generation from trending articles.

🔍 Core Workflows

📰 Current Events Analysis
Multi-source RSS feed collection from tech/business sources with AI analysis, hot topics detection, and relevance boost scoring.
🌐 Enhanced Internet Search
Multi-provider search across DuckDuckGo, Google Custom Search, and Bing with intelligent fallbacks and content extraction.
📊 Advanced Analytics
Week-in-review trend consolidation with D3.js word clouds, keyword frequency analysis, and comprehensive summaries.
👥 Customer Management
CSV-based customer database with cluster organization, reference materials, and intelligent cache management for research insights.

🎙️ AI-Generated Daily Podcasts

Discover our AI-generated podcasts featuring Edge-TTS dual voice synthesis (Alex and Sam personas) discussing trending technology and business topics from current events analysis.

📊 Week in Review Analytics

Explore comprehensive weekly analyses with LLM-powered keyword consolidation, trend analysis, and interactive D3.js word cloud visualizations from 7 days of current events data.

📥 Download MattGPT

📚 Documentation & Guides

Complete documentation to help you get started with MattGPT's features and customization options.

📖 User Guide 🔧 Installation Steps ⚙️ Customization Info 🗺️ Project Schematic

🚀 Installation Instructions

⚠️ Advanced Setup Required - This is a local AI research platform running entirely on your computer

  1. Prerequisites: Python 3.12+, modern web browser, 16GB+ RAM recommended
  2. Download Archive: Click the download button below to get the complete MattGPT source code
  3. Install Ollama: Visit ollama.ai and install for your operating system
  4. Download AI Model: Run ollama pull granite3.2:8b (IBM's Granite 3.2)
  5. Setup Python Environment:
    • python3.12 -m venv .venv
    • source .venv/bin/activate (Mac/Linux) or .venv\Scripts\activate (Windows)
  6. Install Dependencies: pip install -r requirements.txt
  7. Install Additional Tools: FFmpeg for audio processing, optional Edge-TTS for voice synthesis
  8. Configure Sources: Edit sources_config.json to customize RSS feeds and search providers
  9. Launch Application: Run ./MattGPT.sh or python app.py
  10. Access Interface: Open browser to http://localhost:5000

🔧 System Architecture

  • Two-Track Design: Chain Operations + Enhanced Chat separation
  • Local AI: IBM Granite 3.2:8b via Ollama (no cloud dependencies)
  • Modular Components: Content Curator, Current Events Analyzer, Podcast Generator
  • Data Storage: SQLite database with CSV-based customer management
  • Optional Integrations: Google Custom Search, PDF processing, Edge-TTS

🎯 Key Features

  • Privacy-First: All processing happens locally on your machine
  • Customizable: JSON configuration for sources, interests, and analysis parameters
  • Multi-Format Output: Articles, podcasts, word clouds, and analytical reports
  • Customer Insights: CSV-based customer database with research capabilities
📦 Download **NEW** MattGPT Archive (Complete Source Code)

💡 Development Philosophy

MattGPT demonstrates that sophisticated AI research workflows can run entirely locally without cloud dependencies. Use AI assistants like Claude or ChatGPT to help troubleshoot and customize the system, but maintain full control over your data and processing.

🔧 Technical Architecture Overview

MattGPT implements a sophisticated Two-Track Architecture enabling both complex automated workflows and conversational interactions, all powered by local AI models.

🏗️ Two-Track Architecture
Separate operational tracks: Chain Operations Manager for complex workflows, Enhanced Chat Manager for conversational interactions with proper separation of concerns.
🤖 Local AI Integration
IBM Granite 3.2:8b model via Ollama for content analysis, summarization, relevance scoring, and conversational responses without cloud dependencies.
🔄 Chain Operations
Direct orchestration of complex workflows including current events analysis, week review generation, and podcast creation with progress tracking.
💬 Enhanced Chat System
Stateful conversational interface with session management, article context, customer database queries, and URL analysis capabilities.
📊 Content Processing
Multi-source RSS collection, web scraping, PDF processing, and AI-powered content curation with relevance scoring and hot topics detection.
🎙️ Audio Generation
Edge-TTS integration for high-quality voice synthesis with dual speaker personalities and intelligent dialogue script generation.
🏗️

Deep Technical Documentation

Explore comprehensive technical documentation including system architecture diagrams, API endpoints, database schemas, and implementation details for every component.

📖 View Full Architecture Documentation