A major challenge in remote meetings is that awareness cues, such as gaze, become degraded despite playing a crucial role in communication and establishing joint attention. Eye tracking allows overcoming these obstacles by enabling augmentation of remote meetings with gaze information. In this project, we followed a participatory approach by first distributing a scenario-based survey to students (n=79) to uncover their preference of eye-based joint attention support (real-time, retrospective, real-time & retrospective, no) for remote university meetings. Building on these findings, we developed EyeMeet, an eye-based joint attention support system that combines state-of-the-art real-time joint attention support with a retrospective attention feedback for remote meetings. In a four-week study, two student groups worked remotely on course assignments using EyeMeet. Our findings of the study highlight that EyeMeets supports students in staying more focused on the meetings. Complementing real-time joint attention support, retrospective joint attention feedback is recognized to provide valuable support for reflecting and adapting behavior for upcoming meetings.