Ease the access to AI functionalities

Describe the feature you'd like to request

The features provided by the AI integration in nextcloud are not easily accessible and I wish I could use them more easily.

Describe the solution you'd like

I see a couple of interesting features to enhance the integration of AI features. If required, I can split in several Feature requests to discuss them separately, but I found interesting to provide the big picture first.

For me it is especially interesting for the pictures but I thought it could be interesting to consider it globally.
For the pictures, in brain storming, we do not want to click every time: save this file, save this file, etc... and focus on the prompt itself. Eventually, come back to the previous results at the end...

Globally:

• A smart picker should be available and visible in all apps (like the search bar) for a quick access to chatGPT txt, image generation, etc.
• Results should be automatically preserved (in a default folder, configurable by the user):
  • The preservation is actionable with a checkbox (the folder location is prefilled but might be modified), where the AI prompt is shown,
  • If the current app is Files, then an extra checkbox offering to save the results in the current folders,
  • If necessary (or never => configurable), each interaction creates (if not exists) a sub-folder with a templated name (configurable by the user), like ai-pict-YYYYmmdd , translate-de-en-YYYYmmdd, ... with an automatic increment

The following quick actions would be available (similar to what is already available in Deck + what I list below):

• Image generation:

  • Save: Prompt-text in a json file and the result pictures with the same name as the prompt + auto increment per image (or use what is provided in the response, if relevant: I do not know how it looks like)

• Translations:

  • Save: A json file with initial text, its langage and the result (maybe the response from openAi / LocalAi as it comes back)

• ChatGPT:

  • Save: A json file with the prompt and the result (maybe the response from openAi / LocalAi as it comes back)

Describe alternatives you've considered

I did not consider any alternatives.

The issue names no files, tests, or entry points to inspect. Start by breaking the broad proposal into separate feature requests and defining the scope and acceptance criteria for one quick action; completion is not yet bounded by the issue.