Large language models (LLMs) have the potential to support teachers by addressing challenges related to time constraints and data literacy, helping them engage more effectively with student learning analytics. While intelligent tutoring systems (ITS) provide valuable data, many teachers find it challenging to extract important insights for instructional decision-making. This study explores how LLM-generated summaries based on assignment reports can provide concise, actionable insights that improve teachers’ ability to understand reports and respond to student performance. Using a randomized design, we aim to evaluate teachers’ perceived usefulness of summaries, their effect on engagement with reports, and the impact on student participation and outcomes.