What's the difference between Reusable Prompts and Custom Chat Modes

[jjshao-ms]

These two seem quite similar to me. Could someone clarify what the difference is between them?

[aaronpowell]

A Custom Chat Mode is a way to create a new system prompt and define the behaviour of an agent. Here's an example of the PostgreSQL chat mode in which I add an agent that is specifically designed to be my PostgreSQL DBA.

A Reusable Prompt is something you use within an AI assistant workflow, it could be agent or ask, something that you might not want to have to type out frequently because it's a common request you put forth. This is then added to any existing context, rather than changing the system prompt of the interaction.

[anchildress1]

I also have a slightly different write up of both reusable prompts and chat modes on my blog. It could help explain what each does from a beginners perspective. There's one for instructions, too, if you dig around a little. 😉

[falzuabi-sd]

Honestly, I can't find convincing answer in all of the documentation and explanations I see around. I think the makers of these still testing and not sure how to best use them.

Here are the facts about both prompt files and chat modes:

1. Prompt files are appended to the "User Instructions" section in the prompt vs Chat Modes are appended to the "System Instructions" part of the prompt.
2. in Prompt files, you can assign a chat mode, but you can't assign tools or model. in Chat modes, you can assign tools and models.

Beyond these facts, it's up to the user!

on point 1, it's up to the model and the backend to decide how it wants to follow system vs user instructions. I also see this behavior changing as models are advancing/ changing.

on point 2, I see the benefit of using chat modes to assign specific tools and models, but why can't that be added directly to prompt files? maybe it's coming soon?

Great discussion , I came here to ask exact question above and I hope to get better answers or design soon!