Skip to main content

Pin Individual Messages Within a Chat

Feature Request

I often find myself asking follow up questions to a certain answer, which I don’t want to create a separate branch for. Being able to pin a response and get back to it quickly instead of having to scroll up or ctrl+f to find it would be amazing, especially when there are lots of follow ups in long conversations.

Status: Gathering Interest8 comments

Log in to comment and vote

Comments8

  • Moccasin River

    •

    Merged request

    Save individual messages

    I often branch messages multiple times after the same message to not bloat a model with different follow-up questions I have. I want to save often just a single message (which might not be the last one due to follow-up questions afterward again), bloating up the pinned chats and making it currently difficult to find them again. A feature to save individual messages and optionally add a note to them, as well as a menu with all saved messages would greatly help!

    • Mark Florkowski

      Team

      Would you want to discard the remainder of the conversation? Where would you access these saved messages, are you thinking a separate area in the app, or would they be in the sidebar as well?

      I can see the value of what you are describing, but am having a hard time imagining what it would look like from an application perspective

  • Lime Sunbeam

    •

    Merged request

    Flag/Checkpoint Feature

    So I think this is a new and unique feature i haven’t seen in any AI chat app and i think its most crucial.

    An ability to Flag or set checkpoints in a chat thread and a checkpoint nav appears on side to quickly go back to that point in the chat thread. What do you guys think?

  • Magenta Volcano

    •

    Merged request

    Sticky Sections for LLM Prompts/Responses

    I would like to request a feature that allows users to designate a specific section of an LLM prompt or its response as "sticky." This sticky section would remain visible and fixed on the screen while the user scrolls through the rest of the prompt or response.

    Use Case Example:

    Currently, I am studying compiler constuction. When working with complex explanations from an LLM, such as the construction of an LR(1) automaton, it is highly beneficial to be able to continuously compare the LLM's explanation with a reference, like the grammar itself. For example, if I provide the following prompt:

    For the grammar: 

S′ → A0 
A → ABa0 | c1 
B → b0 | ϵ1 

Explain how to construct canonical lr1 automaton, including how we get the lookaheads and the action table. Explain all the follow and first sets used and how they help in coming up with the lookaheads for each rule.

    I would ideally like to keep the grammar (the first part of my prompt) visible at all times while scrolling through the LLM's detailed explanation of the automaton construction, the first and follow sets, and the action table. This would eliminate the need to scroll back and forth repeatedly, significantly improving comprehension and user experience.

    Benefits:

    • Enhanced comprehension of complex LLM outputs.

    • Reduced cognitive load by minimizing constant scrolling.

    • Improved efficiency when comparing different parts of a long response or comparing the response to the initial prompt.

    • Applicable to various domains beyond computer science, such as legal document analysis, scientific explanations, or detailed step-by-step instructions.

    This feature would greatly enhance the usability and effectiveness of interacting with LLMs, especially for educational, analytical, and technical tasks.

  • Peach Leopard

    Exactly this, pinning is essential for UX. t3.chat PLEASE IMPLEMENT THIS. Implement something like Discord’s pinning system. Pin icon top right to view pinned messages. Pin additions and removals could show up in the chat history too maybe (like discord), or show a toast (but i know, toasts bad). You could even create tool calls that the model can use to pin it’s own messages or manage pins. This would also open up the opportunity to implement other tool calls that are specific to using t3.chat itself. But that’s a bit out of scope of the pinning idea.

  • Rose Exoplanet

    constantly thinking about this. will often make unnecessary branches and not touch them so i can always look at the message - stuff like architecture etc

    • Rose Vine

      I’ve been thinking about it more and I think that a canvas feature would work even better instead. If most of what you do with AI is coming up with, refining and iterating on something, having the LLM apply changes to a live document that is always there would work way better.

  • Maroon Spacewalk

    Yes pleeeeease, I was coming here to suggest this, there is a lot of open space on the right side of the chat too atleast on desktop for such a thing.