Skip to content

GuilhermeP96/Aios

 
 

Repository files navigation

🤖 Aios — Local AI Desktop Assistant

On-device intelligence. No cloud. No API keys. No data leaving your machine.

Python Ollama PySide6 Privacy


🔒 Why Aios?

Every AI assistant today sends your data to the cloud — your questions, your context, your private information. Aios doesn't.

It runs a full LLM inference engine locally via Ollama, sits as a floating bubble on your desktop, and responds to natural language queries entirely on your machine.

Built before "local AI" was a buzzword.


✨ Features

Feature Description
🔒 100% Offline All inference runs locally via Ollama — zero network requests
🫧 Floating Bubble UI Always-on-top draggable bubble for instant access
💬 Natural Language Chat Full chat UI built with PySide6/Qt
🧠 Pluggable Brain Modular agent/router architecture — swap LLMs or add skills
Graceful Fallbacks Works even if model or brain is temporarily unavailable
🔧 Configurable Change the LLM model in one line in llm.py

🏗️ Architecture

User Input
    │
    ▼
bubble.py ──► ui.py (Chat UI)
                │
                ▼
           agent.py (Router)
                │
                ▼
           brain.py (Logic + Safeguards)
                │
                ▼
           llm.py (Ollama wrapper)
                │
                ▼
         Local LLM Model
         (runs on your GPU/CPU)

Files:

  • app.py — Entry point, starts the Qt app
  • bubble.py — Always-on-top floating bubble widget
  • ui.py — Main chat interface
  • agent.py — Routes user input to the brain
  • brain.py — Input sanitization + LLM calls
  • llm.py — Thin Ollama wrapper

🚀 Quick Start

Prerequisites

  • Python 3.10+
  • Ollama installed and running
  • Any Ollama model pulled (e.g. ollama pull llama3.2)

Install & Run

git clone https://github.com/harshitgavita-07/Aios.git
cd Aios
pip install -r requirements.txt
python app.py

A floating bubble will appear on your desktop. Click it to open the assistant.


🔮 Why This Matters

2025 is the year of on-device AI:

  • Apple Intelligence runs on-device
  • Meta's LLaMA models run on consumer hardware
  • Privacy regulations are getting stricter globally

Aios was built with this philosophy from day one: your AI, your machine, your data.


🛣️ Roadmap

  • Voice input / TTS output
  • System prompt customization via UI
  • Plugin system for custom skills (web search, calendar, etc.)
  • Hotkey to summon the assistant from anywhere
  • Memory / conversation history persistence

Star ⭐ if you believe AI should stay on your machine.

Built by Harshit Gavita

About

A local, offline AI desktop assistant prototype demonstrating on-device intelligence without cloud dependency.

Resources

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Python 100.0%