Skip to main content
PIO picks an action for each player at decision points you define, then learns from the valued events your game already sends. Tutorial pacing, first-level difficulty, offers: you define the decision points and the candidate actions, and PIO carries the learning loop. Your game asks for a decision, applies the returned action, and keeps playing. The events you track record how that choice worked out, and the reward is derived from them. Models start from a safe baseline and take traffic only as they prove themselves.

Quickstart

Six steps from install to a verified first decision.

Integration guide

Every part of the SDK surface, one topic per page.

Integrate with your AI

Hand the integration to Claude Code or Cursor in one paste.

API reference

The full Unity SDK reference, generated from source.