eriksfunhouse.com
Gimle
Home for my research on dynamical systems — learning their explicit structure from data, representing it so it stays inspectable, and reasoning about it. Many of the other projects here are results of this research.
Hugin
A state machine framework for agentic reasoning. Rather than treating an LLM as the agent, Hugin treats it as an oracle — one component in a larger reasoning system, built around an immutable stack that makes branching, debugging, and multi-agent coordination natural.
Mimir
A foundation model for dynamical systems. Rather than approximating dynamics as a black box, Mimir is trained to discover their explicit, typed, compositional structure directly from observed trajectories — recovering models you can read, compose, and reason about.
Causa
Live dynamical-system models of real-world data, refitted daily. Each model is a causal graph of coupled time series, where every link is labelled by whether it was fitted from data or merely asserted — so you can ask whether a causal relationship is real, and what a shock propagates into.