Occasionally, ideas want to become something more concrete.
An attempt to take equilibrium thinking seriously in software. Instead of asking "what's optimal?", it asks "what's stable?" Instead of maximizing, it watches for convergence.
It grew out of frustration with arbitrary stopping conditions in
iterative systems—for i in range(10) feels like giving up.
What if the system could recognize when it had found balance?
A domain layer on top of TaoCore for processing images and videos into behavioral signals. It extracts what can be observed—faces, poses, proximity, co-occurrence—without claiming to know what people are thinking or feeling.
The system is designed around uncertainty: every signal carries confidence, non-convergence means ambiguity, and refusal is built in. It asks "what patterns are stable?" not "what is this person like?"