I design and build EM-NOUS — a conversational AI governed by an original control architecture that decides how it responds, underneath the language model rather than inside a prompt.

Most systems are stateless: text in, text out, nothing held in between. EM-NOUS is built the other way around — a persistent internal state, memory that survives across conversations, and a layer of restraint that lives beneath the model and shapes its behavior as it happens. The language model is the voice; the architecture is what's steering.

• 🔭 Building EM-NOUS — the architecture itself, deployed and running live. Not a slide deck, not a prototype.
• 🖥️ A desktop client — bringing the system to a downloadable app, kept intentionally small and invite-only for now.
• 🧠 Learning — Rust (for the desktop shell) and the deeper end of control theory.
• 🎯 Focused on keeping the claims honest — the docs should say exactly what the system does, and nothing it doesn't.
• 💬 Ask me about giving an AI a state of its own, memory that means more than keyword matching, or safety that's designed in rather than bolted on.