From Veteran Worlds to New Frontiers: Designing AI Assistants That Learn and Evolve With the Game

We present a robust AI assistant for EVE Online, demonstrating how modern language models can tap into decades of community knowledge already internalized in today’s AI systems. By comparing this with a new, in-development game where such knowledge is absent, we show how our RAG-like feedback loop continuously enriches understanding and delivers a strong player experience in both worlds — each in its own way.
Our architecture emphasizes simplicity and player impact, enabling seamless integration of future features such as recommendation systems built directly on top of the current framework.