PrivaEmanator

Description

A Desktop AI Companion That's Truly Yours

PrivaEmanator chats with you, listens to you, sees your screen, and remembers every conversation you share. All AI runs on your own computer: no cloud, no subscription, and your data never leaves your machine.

⚠️ Before You Buy

All AI runs locally, so response speed depends on your hardware.

• Windows: Requires an NVIDIA RTX 30 / 40 / 50 series GPU

• macOS: Requires Apple Silicon (M1 or later)

Why PrivaEmanator?

• Fully Offline — Works without internet. No one else can see your conversations

• One-Time Purchase — No monthly fees, no token limits

• Your Model, Your Choice — Load community fine-tuned GGUF models from Hugging Face, free from cloud service restrictions

Features

• Chat — Real-time conversation powered by a local LLM

• Agent — Extend your companion's abilities with custom Skills and CLI execution

• Vision — Sees your screen and responds to it; can be toggled on or off in Settings

• Voice Output — Natural-sounding TTS (Chinese and English only)

• Voice Input — Speech-to-text input

• VRM 1.0 Support — Use your own 3D character

• Free Camera — Rotate and view from any angle

• Touch Interaction — Reacts to your touch

• Memory — Remembers your conversations and gets to know you better over time

• Scheduled Tasks — At a set time, send a prompt to your companion or pop up a desktop notification

• Edge Docking — Tucks into the screen edge while you work or watch videos; hover to bring it back

• Multiple Themes — Normal, Scanline, Particle

Make Your Own Characters (VRM 1.0 Characters)

• Supports VRM and VRMA. Add your VRM characters to the assets/vrms folder.

• Custom VRMA animations are also supported: create a folder under assets/vrms/[character]/vrma/ and place your VRMA files inside.

Custom Voices

• Supports TTS ONNX voice models. Drop them into the matching folder under assets to change the voice.

Custom Personality & Memory

• System Prompt, chat history, and memory files are all editable, so it becomes exactly the companion you want

LLM Model Support

• GPU layer offloading (automatically splits model layers between VRAM and RAM)

• Supported models: Qwen3, Qwen3.5, Qwen3.6, Llama 3.2

Agent Skill Support

• Built-in skills: Describe Image, Describe PDF

• Available skills: madato (read XLSX), everything (file search), simple-web-fetch, remider

• Extend your companion's abilities with custom Skills and CLI execution

Scheduled tasks

• At a set time, send a prompt to your companion or pop up a desktop notification

  

Edge Docking

• Tucks into the screen edge while you work or watch videos; hover to bring it back (In games, use borderless windowed mode)

Tutorials

• For step-by-step tutorials on TTS voice models, LLM models, VRM characters and more, check the Guides section in this product's Community Hub.