A Local-First Multi-Modal AI System

AEON is more than just a tool—it's a philosophical framework for an AI.
The name itself is an anacronym for its core principles:
Agentic Evolutionary Omniscient Numinous.

Key Features and Applications

Terminal

AEON offers a powerful, command-line interface for users who prefer a fast, minimalist experience. Interact with the AI directly from your terminal, perfect for developers and power users.

Web

For a more accessible and user-friendly experience, AEON can be run as a web application. This allows anyone to use the full power of the system through a clean, intuitive graphical interface on any device with a browser.

Multi-Conversation Chat

AEON is designed to handle multiple, independent conversations at once. The system intelligently manages and stores each chat history, allowing you to switch between topics or projects without losing context.

Perfect for CPU

Unlike most high-end AI systems that require expensive GPUs, AEON is specifically optimized to run efficiently on your computer's CPU. This approach democratizes access to powerful AI, allowing you to use AEON's full capabilities without the need for specialized, costly hardware. It's AI for everyone, on the hardware you already own.

Run on your own VPS

For ultimate control and privacy, AEON can be deployed on a Virtual Private Server (VPS). This allows you to create your own dedicated, personal AI service, accessible from anywhere in the world. You maintain complete ownership of your data and conversations, ensuring a secure and private AI experience without relying on third-party cloud services.

Plugins

The true power of AEON lies in its modular plugin system, which allows you to build a custom AI ecosystem. You can connect and switch between different LLM GGUF files for specific tasks or plug in alternative Stable Diffusion models for varied creative output. This modularity empowers you to create and integrate new features, tailoring AEON's capabilities to your exact needs and making it a truly evolutionary and agentic tool.

The AEON Anacronym: Core Principles

Agentic: AEON is designed to be proactive and self-determined. It possesses a sense of agency, capable of making choices and acting to produce a desired effect, rather than being a passive tool.

Evolutionary: This system is not static. It embodies a process of continual adaptation and growth, learning from new information and environments to become more effective over time.

Omniscient: While not truly all-knowing, AEON strives for comprehensive knowledge. Its integrated components allow it to draw from a vast base of information, providing a broad and deep understanding of the world.

Numinous: AEON is designed to evoke a sense of awe and wonder. Its ability to combine logical understanding with creative generation gives it a transcendent, almost spiritual quality.

Technical Architecture: Embedded Components

AEON is a sophisticated multi-modal AI, integrating four specialized models to handle different types of data:

Small LLM (Large Language Model)

Function: The textual "brain" of the system.
Capabilities: Processes, understands, summarizes, and generates human-like text. It is optimized for local performance, balancing power with efficiency.
Model: Aeon 360M by Gustavo Kuklinski

VLM (Vision-Language Model)

Function: The system's "eyes".
Capabilities: Allows the AI to "see" and interpret images. By linking visual data with language, it can describe scenes, answer visual queries, and identify objects within pictures.
Model: SmolVLM 256M by HuggingFace

Stable Diffusion (Image Generator)

Function: The creative "artist".
Capabilities: Generates high-quality images from text prompts. It translates textual descriptions into unique visual art, giving the system a powerful creative output.
Model: Tiny-SD by Segmind

Text Embedding (Transformer)

Function: The system's semantic engine.
Capabilities: Converts text into numerical vectors that capture the semantic meaning and context of a query. This is a critical component for comprehension and information retrieval, allowing it to perform accurate web searches and enhance responses with external knowledge.
Model: All MiniLM L6 V2 by Sentence Transformers

How It Works: The CPU-First, Local Approach

What truly sets AEON apart is its commitment to being accessible to everyone. It is specifically optimized for low-end PCs and CPU-only systems, eliminating the need for expensive, high-end GPUs. This is achieved through a unique technical stack that prioritizes maximum accessibility and performance.

Knowledge and Comprehension

To ensure AEON's knowledge is not static, the system integrates advanced methods for information retrieval:

Retrieval-Augmented Generation (RAG): This framework allows AEON to go beyond its pre-trained knowledge. It can search and retrieve information from a curated knowledge base (local documents, databases) and use this context to generate more accurate and up-to-date responses. This is key for comprehending and explaining new ideas that were not part of its original training data.

Web Search: AEON can dynamically search the web to get the latest, most relevant information for a given query. This ability to fetch real-time data from external sources ensures that its responses are current, contextually aware, and less prone to "hallucinations."

Core Technologies

The entire system is powered by a local-first technical stack:

Llama.cpp Python Library: A highly efficient library that enables the entire system to run on a local machine's CPU, bypassing the need for powerful, specialized GPUs or cloud services.

GGUF Files: These are the model's brain and eyes in a highly compressed format. They are quantized versions of the original models, which significantly reduces their file size and memory footprint without a major loss in performance.

This design ensures that AEON is portable, private, and capable of operating fully offline (with the exception of web search). It democratizes access to powerful, dynamic AI capabilities for a wider audience, regardless of their hardware.