AI generated documents follow up

[juliusknorr]

Follow up as discussed with @julien-nc and @elzody for the AI generation (in order of priority):

• [ ] Templates: Make use of a nice looking template files for presentations (e.g. https://github.com/nextcloud/example-files/pull/13)
• [ ] Templates: Have a set of templates in alignment with the design team for different cases that the LLM can decide on and pick one that fits the request best automatically @marcoambrosini
• [ ] Insert images into generated Slides
  • [ ] Blocker: Clarify with Collabora how images can be inserted into the slides (maybe as base64 with existing UNO command
  • [ ] We can ask the LLM to add image generation prompts for slides where it makes sense
  • [ ] Then we can start separate image generation tasks to generate those images and fill them in to the slides
  • [ ] We can also experiment with other slide formats (right now we have bullet lists, but we could have text block, image on the left, image on the right an such)

Out of scope

• Templates: Let the user decide which template to use is rather complex and will be skipped for now

[marcoambrosini]

Can someone add a little bit more context to this? How would this feature work from the user's perspective?

[juliusknorr]

Mostly through assistant: https://cloud.nextcloud.com/s/EbyoFYJQ3xKJRzE

[marcoambrosini]

@elzody what would you need from my side here?