Automated writing evaluation (AWE) provides an instant and cost-effective alternative to human feedback in assessing student writing, and therefore is widely used as a pedagogical supportive tool in writing instruction. However, studies on how students perceive the usage of AWE as a surrogate writing tutor in out-of-class autonomous learning are rare. This study employs a convergent parallel mixed methods approach to explore how students perceive the effects of an AWE program, "iWrite," in autonomous learning context. The subjects of the current study are 146 non-English major undergraduates at a public university in China. The findings indicate that students are overall satisfied with using "iWrite" as a surrogate writing tutor with minimal human facilitation in autonomous learning. They are willing to make repeated revisions to their writing based on the feedback from the automated writing tutor. The results also suggest that the accessibility of AWE tool in out-of-class use could enhance learner autonomy, as students exhibit increased engagement and improved self-regulation following the 16-week intervention. Students perceive "iWrite's" language-based feedback very positively, but their perceptions of content-based feedback from "iWrite" are comparatively negative. Findings have implications for the implementation of AWE in autonomous learning as well as the design of AWE systems in education setting.