Allow users to schedule a chat to send when usage resets. This is useful to the user because 1) they have maximum working memory/context about the task in that moment and 2) they don’t need to remember to return to it. This is useful to OpenAI because it preserves demand that may otherwise be lost and may increase upsell: users can build a small backlog of work waiting on more capacity, making the value of a higher usage tier more obvious. To avoid wasting future compute on low-value requests (those that weren’t memorable enough to return to), the number of schedulable messages could be limited.