buddyGenAI Docs

Appearance

buddyGenAI Options ​

General ​

  • Tab Icons: Whether to display icons for the sidebar tabs. Disable to show text labels instead.
  • Color Mode: Whether to use Light or Dark mode, or use the System-preferred mode.
  • Your Name: This will show up in-chat and control how your buddies refer to you.
  • Profile Image: Set a profile image for yourself to show in chat messages.
  • Your Description: Info about you/the user, which will be provided for extra context when chatting with buddies or the Assistant.

AI Providers ​

At the top of this section, you'll find 4 dropdowns for the features the app supports, allowing you to set which provider handles certain things or Disable them.

These dropdowns and their supported providers are:

  • Chat: KoboldCpp / WebLLM / OpenRouter
  • Image: KoboldCpp / SwarmUI
  • TTS: KoboldCpp
  • Transcription: KoboldCpp

Providers that are selected above will display below in order to configure them. Below is info about how each AI Provider can be configured in this panel:

  • KoboldCpp: Currently you can just set a Host (if you're using something besides the default). Once a valid Host is set, you should see a list of features that the KoboldCpp instance supports.
    • SwarmUI preliminary support is added as an alternative option for local image generation. Like with Kobold, you are able to only set a Host as an option at this time.
  • WebLLM: If enabled, then a dropdown and button will be available to choose and load a chat model (which will be downloaded the first time, should load faster next time)
  • OpenRouter: If enabled, a text box will display to view or edit your connected API key. If there's no API key value in the box, then a "Connect to OpenRouter" button which will take you to https://openrouter.ai to login & get an API key.

Chat ​

  • Stream AI Responses: Control whether messages are typed out as it's received. When disabled, Assistant/Buddy messages will show {Name} is typing... until the entire message is received.

Image ​

  • Enable Chat Images: Experimental feature to allow generating an image for any AI message. Generated images are at 512x768 resolution instead of the square 512px used for profile pictures. When enabled:
    • You'll see an extra option Chat Image Quality which changes how many steps are used to generate images.
      • low = 16 steps
      • medium = 24 steps
      • high = 32 steps
    • You'll now have an image button available on buddy and assistant messages, which will begin loading an image once clicked.
    • This feature is experimental because the resulting prompt/image is generally not the most contextually sensible.

Voice ​

Text-to-Speech ​

  • Default Voice
  • Auto-Read Chat: Once messages are received, automatically generate & play them with TTS.

Transcription ​

  • Auto-Send after recording: Whether to automatically send your input after recording + transcription is finished.