This study explores how AI-supported chess training affects students’ working memory, decision-making, and perceived learning. Conducted over three weeks with students aged 7–12, the research compared an experimental group using AI-guided feedback with a control group playing without AI assistance. Cognitive outcomes were measured using a 2-back memory test, gameplay metrics (accuracy, blunders, and reaction time), and post-training questionnaires. The experimental group showed significant improvement in working memory (p < 0.05), with move accuracy increasing by over 10% and blunders decreased by more than 60%. A strong positive correlation (r = 0.98, p < .01) was found between working memory and strategic decision-making. Additionally, 76.7% of students reported improved thinking, and the average focus score was 4.03 out of 5. These findings suggest that AI-driven chess tools can enhance cognitive performance and self-regulated learning in school settings and offer a valuable complement to traditional instruction.