GREY

Overview

What Is GREY

Local-First

Runs entirely on your machine. No cloud services, no remote servers, no data leaving your network.

Privacy-Focused

You own your data and models. Nothing is transmitted, tracked, or stored externally.

Voice-Enabled

Speak naturally to your AI. Continuous speech recognition with Whisper and Vosk integration.

Leverages Ollama for local model inference. Choose any model that fits your hardware and needs.

No Subscription Required

Free and open source. No recurring fees, no usage limits, no vendor lock-in.

No Cloud Dependency

Works fully offline. Your AI environment remains operational without internet access.

GREY gives you complete ownership of your AI experience. Your conversations, your preferences, your models, and your data all reside on your machine. There are no remote APIs to depend on, no third-party services to trust, and no subscription walls to navigate.

Built for developers, researchers, and privacy-conscious users who believe AI should serve the user, not the other way around. GREY transforms your local machine into a capable AI operating environment with voice, memory, tools, and autonomous agent workflows.

Capabilities

Key Features

A comprehensive local AI environment with voice, memory, tools, agents, and system monitoring in a single application.

Voice Interaction

Continuous speech recognition
Whisper integration
Vosk support

Local AI Models

Ollama integration
Model flexibility
Local inference

Persistent Memory

User preferences
Project tracking
Long-term context

Tool Execution

File management
Browser automation
System controls
Search

Agent Workflows

Multi-step planning
Task execution
Error recovery

System HUD

CPU monitoring
Memory monitoring
Network metrics
Real-time status

Drag and Drop Processing

PDFs
Images
Code
Documents

Settings Management

Configure everything from UI
Model switching
Voice selection

Comparison

Why GREY

When you choose local AI, you choose control, privacy, and independence. See how GREY compares to cloud-dependent alternatives.

Capability	GREY	Cloud AI
Privacy	Full — data never leaves your machine	Data sent to remote servers
Offline Capability	Works without internet	Requires constant connection
Local Models	Run any Ollama model locally	Limited to provider models
Subscription	No subscription — free and open	Monthly fees required
Full Control	You control everything	Provider controls your experience
Local Memory	Persistent local storage	Stored on remote servers
Data Ownership	You own all data	Provider may access or use data
Customization	Full model and voice selection	Limited to provider options

Process

How It Works

Four steps to transform your machine into a local AI operating environment.

Install Ollama

Download and install Ollama for your platform. This provides the local model runtime.

Download a Model

Pull a model like llama3.2 using Ollama. Choose based on your hardware capabilities.

Launch GREY

Clone the repository, install dependencies, and run the application on your machine.

Talk Naturally

Speak or type to interact with your local AI. Use voice, tools, memory, and agents.

Setup

Installation Guide

Get GREY running on your machine in minutes. Follow the steps for your operating system.

Python 3.10+

Ollama

Microphone

                            $ git clone https://github.com/greyisgreat/grey-mark-xl.git

$ cd grey-mark-xl

$ python -m venv .venv

$ .venv\Scripts\activate

$ pip install -r requirements.txt

$ playwright install

$ ollama pull llama3.2

$ python main.py
                        
                            $ git clone https://github.com/greyisgreat/grey-mark-xl.git

$ cd grey-mark-xl

$ python3 -m venv .venv

$ source .venv/bin/activate

$ pip install -r requirements.txt

$ playwright install

$ ollama pull llama3.2

$ python main.py

Configuration

First-Time Configuration

Recommended fully local setup for complete privacy and offline capability.

Speech-to-Text

Whisper

Engine Whisper

Model base

Local LLM

Ollama

URL localhost:11434

Model llama3.2

Text-to-Speech

Kokoro

Engine Kokoro

Type Local TTS

This configuration provides a completely local AI stack — speech recognition, language model inference, and text-to-speech all run on your machine with zero cloud dependencies.

Support

Frequently Asked Questions

Common questions about GREY, its capabilities, and requirements.

What models work with GREY?

GREY works with any model supported by Ollama. This includes llama3.2, mistral, phi3, gemma, codellama, and many others. You can pull and switch between models freely based on your hardware capabilities and use case. Larger models require more RAM and VRAM, while smaller models like phi3 can run efficiently on modest hardware.

Does GREY require OpenAI?

No. GREY does not require OpenAI or any cloud API. It is designed to run entirely locally using Ollama for model inference. While you can optionally configure cloud APIs if you choose, the default and recommended setup is fully local with zero external dependencies.

Can GREY run offline?

Yes. Once models are downloaded and the application is set up, GREY runs completely offline. Speech recognition, language model inference, text-to-speech, memory, and tool execution all function without an internet connection. This is one of GREY's core design principles.

Does GREY store memory locally?

Yes. All memory, preferences, and conversation data are stored locally on your machine. GREY uses local file-based storage for persistent memory, user preferences, and project tracking. No data is ever sent to remote servers unless you explicitly configure a cloud-based model provider.

What operating systems are supported?

GREY supports Windows, macOS, and Linux. The installation process is straightforward on all platforms — clone the repository, create a virtual environment, install dependencies, and run. Ollama, the underlying model runtime, is also available across all three platforms.

Can I use my own Ollama models?

Absolutely. GREY is designed to work with any Ollama-compatible model. You can pull models from the Ollama library, use custom Modelfile configurations, or even serve locally fine-tuned models. Simply configure the model name in GREY's settings, and it will connect to your local Ollama instance running that model.

Roadmap

Future Vision

GREY is evolving beyond a chatbot into a complete AI operating layer for the desktop.

Voice interaction with natural conversation flow

Tool usage for real-world task execution

Persistent memory across sessions and projects

Agent workflows for multi-step autonomous operations

Personal productivity augmentation

Local intelligence that learns and adapts

The long-term goal of GREY is to become a complete local AI operating environment — not just a conversational interface, but an intelligent layer that sits between you and your desktop. It will understand your workflows, remember your preferences, execute tasks on your behalf, and provide persistent, private intelligence that grows more useful over time.

GREY represents a shift from cloud-dependent AI services to sovereign computing — where your AI is as private and personal as your own thoughts, and as capable as your machine allows.