Follow the decisions behind Battle AI: how it chooses an action, how it is tested, and how regret guides a bounded policy change.
Each system has a flowchart, an equivalent text explanation and a full-resolution PNG. These explain the method and its limits; they are not a promise of perfect play.
PokeSynergy starts only from what the acting side can know, scores every legal joint action against plausible opponent actions, spends extra simulation only on tactically distinct close decisions, then returns a deterministic choice with an auditable score breakdown.
Explore system →
PokeSynergy tests tactics and team matchups on separate, complementary axes, promotes only after bounded reproducible evidence passes, and sends failures back into the next candidate rather than hiding them.
Explore system →
Turn and preview regret expose different decision errors, bounded candidate changes rerun the full visible evidence, and only a passing candidate may cross the sealed promotion gate.
Explore system →Looking for practical team-building references? Read the guides, or open Battle Lab to use your saved teams.
