AI Playtest Studio

AI Playtest Studio is a playtesting lab for game managers and analysts. Instead of waiting weeks for real players, a game manager can ask an AI team to play the game and answer two questions: "Where do new players get stuck?" and "Would this new feature actually help?" Find Bugs: persona bots (an impatient newcomer, a story-guided casual player, an optimiser) play a browser city-builder through the real screen. SAM 3.1 marks what is on screen, Muse Spark decides each click as that persona, and the bots think out loud. The run ends with a report that shows the stuck step, the evidence frames, and a proposed config fix. A/B Test a Feature: the manager types a feature idea in plain English (e.g. a Harvest Festival that turns food into score). Muse Wonder turns it into a validated config patch, Muse Image draws its emblem, and it ships when he approves it. The same personas then play both variants on paired seeds, and the report gives a measured verdict (positive, mixed or hurts) with per-persona results and, if needed, a reworked version to test next. Everything runs on schema-validated JSON patches with a human approval gate, deterministic seeds, full telemetry and replayable runs. The game and all assets are original and were built during the event.