[FEAT]: Features and QoL recommendations by my personal opinion

[DanB1172]

What would you like to see?

Settings

• LLM:
  Built-in huggingface model browser and downloader without manually inputting url
• Vector Database:
  Option to import custom DB models(since local one exists by default and the ability to use cloud provided services are also an option)
• Voice and speech:
  1. Add option to download any STT model user wants, not just be stuck with the default( this can be implemented either through the user importing their own file or through a service like hugging face)
  2. should be able to set STT model and all of its settings of meeting assistant in this settings menu instead of meeting assistant its own preference menu.
• Meeting Assistant:
  1. Please move it out of the settings menu and into the main ui.
  2. option to edit templates by right clicking them instead of selecting them and then be able to edit them.
  3. When Starting new meeting, either have the ability to select which template to use before importing a file, or be able to select default template somewhere in settings
  4. Be able to import/choose more STT models other than parakeet in the preference menu or Voice and Speech settings menu

Main Interface/Other

• Allow attachment support for audio files(some multimodels are able to analyze these)
• Change LLM Selection Icon, the brain icon makes it look like its an option for thinking mode or something similar
• I feel like the text size icon shouldnt be in the selction bar below the message input prompts, i feel like placing it in the ui settings menu would make more sense.

[timothycarambat]

LLM:
Built-in huggingface model browser and downloader without manually inputting url

👍

Vector Database:
Option to import custom DB models(since local one exists by default and the ability to use cloud provided services are also an option)

We have plenty of cloud vector databases and Embedder options for providers/models - can you elaborate more on the request here?

Voice and speech:
Add option to download any STT model user wants, not just be stuck with the default( this can be implemented either through the user importing their own file or through a service like hugging face)
should be able to set STT model and all of its settings of meeting assistant in this settings menu instead of meeting assistant its own preference menu.

#5268 for STT to build it out more for more models

Meeting Assistant:
Please move it out of the settings menu and into the main ui.
option to edit templates by right clicking them instead of selecting them and then be able to edit them.
When Starting new meeting, either have the ability to select which template to use before importing a file, or be able to select default template somewhere in settings
Be able to import/choose more STT models other than parakeet in the preference menu or Voice and Speech settings menu

Please move it out of the settings menu and into the main ui.

Where would this go ideally? The main UI is chat so what would this look like?

When Starting new meeting, either have the ability to select which template to use before importing a file, or be able to select default template somewhere in settings

I agree with this!

Allow attachment support for audio files(some multimodels are able to analyze these)

Most models cannot, but we do have on-device transcription already so we plan to have this be supported using that pipeline. The same thing that powers the meeting assistant.

Change LLM Selection Icon, the brain icon makes it look like its an option for thinking mode or something similar

This was addressed in the new UI uplift since this issue was posted. Model selection is now in top-left!

I feel like the text size icon shouldnt be in the selction bar below the message input prompts, i feel like placing it in the ui settings menu would make more sense.

Also changed in the new UI - now in the top right "Settings" icon - where some more options will be soon.