QuerySynth — Hybrid AI+RAG Assistant (Tkinter Desktop)

QuerySynth is a local/online retrieval-augmented AI assistant (RAG) with a powerful Tkinter-based desktop frontend.
It combines chat from a local Large Language Model (LLM) (via Ollama) with live search and retrieval of facts, code, images, audio, translations, and more using real web APIs and knowledge sources.

🚀 Features

• Hybrid Offline/Online Chat: Chat with a local LLM or answer with live data.
• Multiple Modes: "Chat", "Search", and "AI+RAG" (real-world grounded QA).
• Rich Natural Language Retrieval:
  • Wikipedia summaries & articles
  • News aggregation (NewsAPI, Guardian, GNews, NYTimes)
  • Pexels image search
  • YouTube audio download & playback
  • StackExchange code and Q&A
  • Government data, World Bank info
  • Wolfram Alpha queries (math, science, stats)
  • Weather lookups
  • Code snippet/algorithm search
  • Definitions, translation (multi-lingual), and more
• Tkinter GUI: Markdown rendering, clickable sources/links, image display, integrated audio player
• Automatic Conversation History
• One-click Chat Reset: "Refresh Chat" button for quick conversation reset

🖼️ Demo

• 

📦 Installation

1. Clone the repository:

   git clone https://github.com/YOUR_USERNAME/querysynth.git
   cd querysynth

2. Install dependencies (Python 3.8+ recommended):

   pip install -r requirements.txt

3. Install Ollama for LLM local chat:

   • Download from ollama.com/download and install per your OS.

4. Configure API keys (see below):

🔑 API Key Setup

To use all features, API keys are required for several external services.

One Time Setup:

1. Copy the example config file:

   cp config.example.json config.json

2. Register for free developer keys using the following links, paste them into your config.json:

Key	Where to get/register
news_api_key	NewsAPI.org
guardian_api_key	The Guardian Open Platform
pexels_api_key	Pexels API
gnews_api_key	GNews API
weather_api_key	WeatherAPI.com
wolfram_app_id	Wolfram Alpha

config.json should look like:

{
  "news_api_key": "your_news_api_key_here",
  "guardian_api_key": "your_guardian_api_key_here",
  "pexels_api_key": "your_pexels_api_key_here",
  "gnews_api_key": "your_gnews_api_key_here",
  "weather_api_key": "your_weather_api_key_here",
  "wolfram_app_id": "your_wolfram_app_id_here"
}

Do NOT upload your filled-out config.json! Only share config.example.json.

⚡️ How to Use the Frontend

1. Start Ollama server:
   (ensure a supported model like llama2 or mistral is downloaded in Ollama)

   ollama serve

2. Launch QuerySynth:

   python ai_cnv.py

3. Interact via the GUI:

   • Enter your question and press "Send" (or press Enter).
   • Switch between Chat, Search, or AI+RAG (retrieval + LLM) modes.
   • View sources, formatted answers, code, images, and audio in the right panel.
   • Double-click any link to open in your web browser.
   • Use the 🔄 Refresh Chat button any time to clear the chat/session.

4. Conversation history is saved in conversation_history.json (auto-managed).

📚 Technologies & Sources

• Languages/Frameworks: Python 3.x, Tkinter
• NLP: NLTK
• LLM Backend: Ollama (local server)
• APIs: NewsAPI, Guardian, GNews, NYTimes, Pexels, WeatherAPI, WolframAlpha
• Data: Wikipedia, StackAPI, World Bank, Government open data, more
• Python Libraries: Pillow, pygame, requests, beautifulsoup4, markdown2, tkhtmlview, yt-dlp, StackAPI

🛠️ Requirements

• Python 3.8+
• Dependencies from requirements.txt (see file for details).
  • E.g.: nltk, requests, markdown2, tkhtmlview, pygame, pillow, yt-dlp, beautifulsoup4, StackAPI, etc.

Install all requirements via:

pip install -r requirements.txt

🙋 FAQ/Troubleshooting

• Nothing happens when I click Send?

  • Ensure Ollama is running and a model is ready (ollama list).
  • Check your API keys are set and valid.

• API error: invalid key?

  • Make sure you copied each API key exactly and that your account is active.

• Where is conversation history stored?

  • All saved locally to conversation_history.json in your project directory.

• How do I reset/clear chat?

  • Click the "🔄 Refresh Chat" button.

📝 License

MIT License — Free for research, interviews, and educational use.
Always attribute sources if you reuse or redistribute this codebase.

Built and maintained with ❤️ for open AI tools and knowledge search.