Yeah, that’s one flow. The important part is that you accept events as inputs so if I create a 5250 with

[ i, “event-id”, “event” ]

You fetch event-id and use that content as the text to be processed

That way i can:

* create a nostr content discovery event

-> output of that is 50 or 100 e tags of interesting events

* create a summarization job that takes those 50 or 100 events as input and creates a single summarization content

* create a 5250 that reads me the summarization in guy swann’s voice