I’m building a local, symbolic AI assistant without an LLM – and it runs 24/7

Hi everyone,

I'd like to share my project Nova AI with you. It's a personal AI companion that runs entirely locally and uses no large language model. Instead, she's built on symbolic AI: a network of explicit concepts, relationships, and patterns that I can inspect and modify myself.

What it can do right now:

· Have natural conversations

· Play chess against Stockfish (with a colored board and statistics)

· Fetch multi-day weather forecasts

· Query Wikipedia and automatically learn new concepts

· Build word associations using PMI scoring

· Recognize behavioral patterns (timing, frequency)

· Restart herself without losing data

· Her own personality, emotions, and expression

Architecture: EventBus + 7-layer memory (SQLite, associative network, pattern recognition, semantic reasoning, response generation, context, and personality).

The kicker: I'm a self-taught developer from Belgium. A year ago, I couldn't write a single line of code. Everything was built with AI assistance, but the vision and design choices are entirely mine.

The code is public, but the repository is mainly a look behind the scenes – not a plug-and-play package. I wanted to show what's possible when you think outside the LLM hype.

I'm very curious about your questions and feedback!