Despite poetry’s important role in improving linguistic skills and creative thinking, students often find poetry writing to be difficult and boring. This study is an investigation of how the digital Poetry Machine influences students’ poetry writing by applying AI techniques. It uses qualitative and quantitative analysis of the log data of poems that the seventh graders wrote with the Poetry Machine. The results show that the draft poems functioned as affordances, which the students followed as models. The drafts encouraged students to experiment with several different poetic features. The data suggest an association between the number of edited versions and the quality of the final poem. The results suggest that a co-creative AI-based tool inspires and supports those students who engage in the writing process, and the poems are developed from the first versions. More studies regarding the role of AI based digital tools in developing students’ writing competencies would be worthwhile. •A co-creative AI-based tool inspires and supports students who engage in the poetry writing process.•The draft poems produced by the AI-based Poetry Machine function as affordances, which the students follow as models.•The students need more opportunities to use of AI-based tools for writing at school.