As a developer your service to upload files is unsuable with code source files.

This is a terrible lack of functionnality for many LLM provider that allow the upload of source files directly instead of many copy paste into chat

If you upload a .pdf with pictures, then it scans or OCRs the picture to get all the data out for summurization

Possibility of linking the account to folders in the cloud and in particular Google drive for chat docs.

On the pdf chat screen, I want the pdf zoom setting to be 100% by default. But every time I change it it is displayed as 330%.

From a UX perspective I think it would be better if the Select File option in MultiDoc chat was expanded by default. Currently we have to click on the Selected 0 files: 0 words panel and then click again on the select file list to chose files. This seems like one too many clicks. I know it's only a small thing but I think this would make the user experience more fluid.