# heyDot-Windows 🎙️👁️

An omnipresent, voice-first, context-aware desktop AI assistant built specifically for Windows. Triggered instantly by a global shortcut, heyDot captures your active desktop context, transcribes your spoken prompt completely in the cloud, and utilizes advanced multimodal vision nodes to answer queries about what you are looking at—speaking the response back to you natively.

Optimized into a decoupled **Client-Server Architecture**, the desktop environment remains incredibly lightweight, routing all intensive operations through high-speed cloud pipelines.

---

## ✨ Features

* **Global Hotkey Integration:** Tap Alt + D globally from any application or workspace window to instantly wake the assistant.

* **Multimodal Visual Context:** Automatically grabs primary display frame buffers upon trigger—no copy-pasting code or layouts required.

* **High-Speed Cloud Inference:** Powered by Groq Cloud Nodes hosting Whisper-Large-v3 for speech extraction and Llama-3.2-11b-Vision for image reasoning.

* **Native Text-to-Speech:** Leverages the local Windows SAPI5 voice engine to speak responses fluidly.

* **Ultra-Lightweight Client:** Compiles to a clean standalone executable (.exe) with zero local heavy model requirements.

---

## 🏗️ Repository Architecture

heyDot-Windows-Product/

│

├── backend/            # FastAPI Cloud Engine Code

│   ├── server.py       # API endpoints processing Whisper/Llama payload

│   └── requirements.txt# Backend package dependencies

│

├── client/             # CustomTkinter Desktop Interface

│   └── heyDot.py       # Client controller logic \& hotkey hooks

│

└── README.md           # Documentation



\---



\## 💖 Acknowledgements \& Inspiration

This project is inspired by and built upon the core concepts introduced by \*\*\[Shub3am/heyDot](https://github.com/Shub3am/heyDot)\*\*. Huge credit goes to the original creator for the foundational inspiration behind building a context-aware companion workflow!