Hi OpenAI team,
I noticed the new ChatGPT “Create note” feature, where a user can select text in a conversation, click Create note, and ChatGPT creates a writing block at the end of the chat.
This is a useful feature, but I think there is an important UX limitation: the note becomes detached from the original text or code that caused it.
At the moment, the workflow feels like this:
- Select text or code in the chat.
- Click Create note.
- A note/writing block is created at the end of the conversation.
- Later, if I want to find the note again, I need to scroll down or search for it.
This works for capturing notes, but it is weak for retrieval and review because the note loses its original context.
A better mental model might be closer to comments in Microsoft Word or Google Docs. When I select text and add a comment, the comment is anchored to that exact text. I can see the highlighted text, open the related comment, and move through comments in a side panel.
For ChatGPT, I think the ideal behaviour would be:
-
Anchored notes
When I create a note from selected text, the original text/code should get a subtle marker or highlight. Clicking that marker should reveal the note. -
Side notes panel
A right-side panel could list all notes in the conversation. Clicking a note would jump back to the original message, paragraph, or code block that created it. -
Optional writing block output
The current “writing block at the end of the chat” could still be useful when users want to turn notes into a document, but it should not be the only storage/retrieval model.
This matters especially for code and technical workflows. If I create a note from test_insight_generation.py, a function name, an error message, or a specific review comment, the note should remain linked to that exact source. Otherwise, it becomes another detached artifact that I need to manually rediscover later.
In short: the current feature captures notes, but it does not preserve enough location/context. An anchored comment model, similar to Word or Google Docs, would make notes much more useful for review, debugging, coding, and long technical conversations.
Thanks for considering this feedback.
