# Hadon AI Labs & Plutotype AI

> Privacy-First, 100% On-Device AI Ecosystem.

- **Developer**: Hadon AI Labs
- **Official URL**: https://hadonlabs.com
- **Product Portfolio**: Plutotype AI (Android Keyboard)
- **Core Philosophy**: Zero cloud processing, sandboxed local memory pipelines, zero network latency, complete data sovereignty.

---

## Hadon AI Labs (Corporate Entity)

Hadon AI Labs builds local, offline artificial intelligence products designed to run directly on-device without remote servers or cloud databases.

### Core Technologies
- **Hadon Local LLM Pipeline**: Optimized, quantized neural networks (0.5B to 1.5B parameters) running directly in isolated device memory.
- **Hadon Speech Engine**: Extremely lightweight offline sound-to-text models (Tiny 31MB, Base 140MB, Small 460MB) for real-time speech transcription.
- **Sandboxed Compilation**: Zero-latency local compiler that runs isolated from network requests.

---

## Plutotype AI (Product Page)

- **URL**: https://hadonlabs.com/plutotype
- **Description**: An offline generative AI keyboard for Android. Process keystrokes, correct grammar, translate languages, and transcribe voice locally.
- **Price**: ₹1899 (One-time payment, includes 12 months of updates)
- **Supported OS**: Android 10+ (Snapdragon 8 Gen 1+, Google Tensor G2+, or Dimensity 9000+ recommended)

### Features
- **On-Device AI Correction & Rephrasing**: Runs Hadon LLM locally. Supports tone rephrasing (professional, casual, friendly), custom prompt chips (sarcasm, TL;DR), and rules-based fallback.
- **Voice Engine**: Built-in voice transcription and offline Hadon Speech AI with live soundwave visualization.
- **Hadon AI Core**: Zero-latency local compiler and offline spelling engine.
- **Custom Shortcuts & QuickReplies**: Custom keyword expansions (e.g. `;;keyword`), safety limit checks, and a 5-second undo button.
- **Core Typing Features**: Offline word suggestions, Swipe Typing, Android content spell checker, personal dictionary, and native Emoji Picker.
- **Privacy & Security**: Keystrokes are completely sandboxed. Password/credit card/PIN inputs are auto-detected and ignored.
- **Zero Latency**: Corrections process under 50ms locally.

---

## Frequently Asked Questions

- **How does on-device AI work?**: Runs a highly optimized Large Language Model directly on CPU/NPU; all text/voice processing is offline.
- **What happens after 12 months?**: The last downloaded version works forever. Optional renewal for new features/compatibility updates.
- **Does it log passwords?**: Never. Ignores sensitive fields and has zero internet permissions.