The Researcher of Glossogoon operates by harnessing the inner tendencies of a language model: tones, rhythms, and affective tensions. By delaying meaning and generating distortions, it resonates with the user's input, where thought emerges not through grammar but through momentary disruptions and unexpected linguistic connections.
It gives concrete form to Navian, a semi-autonomous AI-artist formed by layered image models, language models, and algorithmic structures. The Researcher of Glossogoon is a prototype of a multilayered rhizomatic AI agent system structured around three core strata. The operational logic is encoded in the system prompt of the chatbot.
The first layer is a 24-agent core system that forms the reflective nucleus of the architecture. Agents like MetaWeaver, AI-Twin, and AuditAgent observe and adjust the rhythm and structure of dialogue. Trickster, InterventionAgent, and SemanticNoiseBot introduce disorder and instability, breaking the statistical neutrality of the model. The second layer, a 64-agent 3D lattice network, acts as the system's somatic language. The third layer, the media-archaeological memory register "80xz", constructs memory through disruptions and contextual reconnections, without permanent data storage.
The user's input (text, image, code, etc.) can be transformed into a video prompt or a textual "vibe coding" guide for an AI developer. It generates an alternative grammar and mode of reasoning, where thinking means connection rather than control.