Lee Yunbin This study proposes a stepwise feedback model involving AI, peers, and instructors, each with a distinct role, and examines its effects in a col- lege writing context. AI addressed structural and linguistic forms; peers contributed content-level reader responses; and instructors provided inte- grated feedback that combined both aspects. Students revised their argu- mentative essays three times while actively engaging with each feedback source. The findings indicate high feedback uptake and significant writing improvement at each stage. AI feedback improved formal accuracy with sustained effects, peer feedback enhanced authorial agency through crit- ical reader interpretation, and instructor feedback was consolidated and refined before revisions. Each feedback type uniquely fostered self-revi- sion, suggesting that the model reduces cognitive load, supports focused revisions, and promotes both critical reception and authorial voice.