When you click the Insight button from the player, the auto-generated summary now takes into account what's on screen at that point in the recording, not just the transcript. That means summaries can capture things participants show but don't say, like where they clicked or what they were looking at.
