Enhancement: The ability to route chats or specific messages to cloud storage targets.

[danielrosehill]

Feature Outline

During the course of interacting with large language models, I frequently find myself wishing that there were the ability to more easily save either whole chats, my prompts, or certain outputs into defined targets.

For example, I'd love to be able to run commands like:

"Save this chat to my Google Drive, folder Y" (Google Drive API)
"Save this chat into my Nuclino" (Nuclino API)
"Save. Just the last output into Google Drive."

By commands, I mean perhaps real "commands" down the line (programmatic instructions delivered within the chat). But, as a more basic implementation, elements would be helpful.

Feature Requirements

Given that there are a zillion APIs out there, it would probably be much more feasible just to focus on the most common storage targets. Thinking about a mixture of business and personal use, I'd suggest these as:

File storage and buckets

• Google Drive
• Dropbox
• S3
• B2
• Confluence

Databases

• MongoDB (I know you already do)
• Postgres
• MySQL

Perhaps even give users the ability to selectively export parts of the chat to an external Weaviate or Pinecone database for selective embedding.

Feature Variations

Some premutations on this idea:

• Give users the ability to choose between automatic saving to their target destinations or more selective saving just for specific aspects
• Allow users to configure a default folder where the conversations or chats are routed to.

Use cases that could be enabled

• In the business context, the user has a chat with a large language model, they're working on an agenda for a forthcoming meeting. After a few iterations with the AI tool, they arrive at the finished product. Now they want to save that to their Google Drive. Complementary agentic capability that would be great too would be the ability to email, but that's a different feature.

• User could configure a specific drive folder as their prompt library. LibreChat could remain the centerpiece for their actual interaction with theme models but by storing the prompt externally it could be more collaborative.

Two-way integration With Cloud Storage Targets

2-way integrations would bring an even wider amount of features.

If the user could connect their Google Drive, they could use it both for saving outputs and also for importing docs as context. This could be very fluid with the user both bringing in and writing out data all within the chatbot window.

Potential Add-Ons

Allow the user to set up actions which are predefined playbooks that the user could populate from a shared library:

"Save just the prompt to my B2 prompts bucket"
"Save the whole conversation to my S3"
"Save just the last output to my Google Drive"

Etc