[IDEA]: Self-Hosted AI Chat Responses (Ollama)

[cotton105]

In the spirit of self-hosted tools, it would be great if the /chat command could support offline AI text generation, for example with a local Ollama instance.

At the moment, it currently seems restricted to using online tools that use an API token (OpenAI, Gemini, Anthropic). If it was possible to query a locally hosted alternative, it would help to keep everything self-contained, and give the Bot Hoster the freedom to choose whichever model they want.

I've done a small amount of research and found an Ollama API for Node, which might help an implementation on Bastion.

[atlanna]

Thank you for opening this issue.
A maintainer will get by as soon as practical to address this issue.

If this is a support question and not really an issue or suggestion, then please ask it in our Discord Server instead.