The post-Attention
AI architecture

We are currently developing and studying the Geometric Potential Model (GPM) architecture.

GPMs are deterministic, traceable, causally auditable, and truly steerable. These characteristics open new possibilities for better, safer and cheaper AI systems.