Request:
Please preserve the user-document's chatbot state for at least an hour since last touched (if not 24 hours). This will allow Summarizer users to use the document chatbot as a blackboard to explore ideas and topics without concern for the Q+A transcript to disappear prematurely.
Configuration:
- OS: macOS 15.2
- Browser: Safari 18.2
What does my feature entail?
When using the Summarizer on a given document (especially a YouTube transcript), it would be awesome to be able to switch back-and-forth between the "Key Moments/Summary" landing page and the "Discuss Further" chatbot about the document.
For example, using the Summarizer on the following Joe Rogan video works amazing:
!sum https://www.youtube.com/watch?v=r_3yOQl9E4I
And gets you the following landing page:
And when I want to go to chat with it, I can do that just fine. Here's a Q+A with the document chatbot:
However, when I want to switch back to the original Summary/Key Moments document, the "Discuss Further" chatbot state is flushed and I have to create a brand new chat session, instead of using the Q+A to fully flesh out the dialogue:
What is it for?
This enhancement will allow Summarizer users to use the document chatbot as a blackboard to explore ideas and topics without concern for the Q+A transcript to disappear prematurely.
How will it affect existing workflows or user experience?
This enhancement will empower users to spend more time with the document chatbot by being able to go back-and-forth between the summarizer document landing page (Started with the !sum bang) and the document Q+A. It will empower the users to be able to use both the "left-hand side" of the Summarizer's brain (the raw summary) and the "right-hand side" (the creative Q+A that a user queries the document chatbot).
I envision users using this enhanced state feature as a way to deep-dive into any kind of summary document and use it as an functional exploration tool.
I'm not aware of any other browser or apps that do something exactly like this proposed feature. I am aware than RAG-based LLMs that are NOT as tuned to be explicitly summary-versatile but do have thread-preservation as a substitute.
If this feature was added, how would it work into it to extend usefulness?
As expressed above, it would make the document chatbot WAY more state preservation-friendly and allow a fuller and deeper exploration of any given document. This would unleash a user's research creativity knowing they can query the document with abandon and still be able to switch back-and-forth to and from the summary landing page.
I think if the chatbot state for a given user-document was preserved for say 6 hours, that would be plenty of time for a user to get their stuff done and possibly export the chat transcript.
See REQUEST: Export Summary Chatbot transcript to Markdown for a discussion about export functionality.
