For years, the headlines in artificial intelligence have focused on “superhuman” systems that “solve” games. These projects are landmark achievements in computation. But in poker, we felt they were solving the wrong problem.

They proved that a machine could master the mathematics of poker in a sterile, heads-up, rake-free environment. They had built a perfect shield. But they had ignored the game itself—the chaotic, multi-handed, psychological contest of wills, intuition, and emotion.

SpiderDime Systems was founded to ask a heretical question: What if the GTO-based “solver” approach isn’t the future, but a dead end?

We believed the true challenge was not to build a better calculator, but a better observer. We set out to prove that in a game defined by human imperfection, the winning strategy is not to be unexploitable, but to be maximally exploitative. This required a new approach. We undertook a focused, multi-year research project to build an AI on a different foundation: not on GTO, but on heuristics, adaptation, and a novel learning model that anchors on predictive accuracy, not just financial results.

Patrick is the result. It is an AI designed to thrive in the chaos of a real game. It doesn’t just calculate; it observes, it profiles, it adapts, and it attacks the flaws it finds.

The assertion that poker is “solved” is a fundamental misunderstanding of the game. On this site, we present our counter-argument: the white paper, the complete hand histories, and the video evidence of our successful trial.