Give the chat basic context #54

Open
opened 2026-11-18 00:58:03 +00:00 by jbwexler · 0 comments

What's the idea?

Currently, the model seems to have no idea about where it's being run, that it's part of inference.coop, that it's using librechat, etc. I think this can be confusing for a new non-technical user who expects to be able to ask the chat questions about how to use it's features and about the co-op more generally. They may eventually figure out they can give the chat access to inference-docs, but that's not immediately intuitive from the interface.

Why does it matter?

I think it's important to make the chat as intuitive and user-friendly as possible to be inclusive for non-technical users.

Who is it for?

New users

How might it work?

I see two possibilities:

  1. Simply give the model access to inference-docs by default. This would be ideal from a user-friendliness perspective, but I have noticed the model will access these docs more readily than it should. For example, when I asked it about the differences between chatgpt and claude, it accessed the docs. Thus, the downside could be excess token usage (maybe this is a minor increase, I'm not sure).

  2. We could include a short default system prompt that tells the model where it's being run and to instruct the user to turn on the inference-docs tool if they seem to be asking questions about the coop or chat.

## What's the idea? Currently, the model seems to have no idea about where it's being run, that it's part of inference.coop, that it's using librechat, etc. I think this can be confusing for a new non-technical user who expects to be able to ask the chat questions about how to use it's features and about the co-op more generally. They may eventually figure out they can give the chat access to inference-docs, but that's not immediately intuitive from the interface. ## Why does it matter? I think it's important to make the chat as intuitive and user-friendly as possible to be inclusive for non-technical users. ## Who is it for? New users ## How might it work? I see two possibilities: 1) Simply give the model access to inference-docs by default. This would be ideal from a user-friendliness perspective, but I have noticed the model will access these docs more readily than it should. For example, when I asked it about the differences between chatgpt and claude, it accessed the docs. Thus, the downside could be excess token usage (maybe this is a minor increase, I'm not sure). 2) We could include a short default system prompt that tells the model where it's being run and to instruct the user to turn on the inference-docs tool if they seem to be asking questions about the coop or chat.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: co-op/support#54