Multimodal Input Support (Images, PDFs, CSVs) in Vanna

[dogra1598]

Does vanna plan to support multimodal inputs such as images, PDFs, or CSV files as part of the query context, instead of only accepting plain text input?
We are evaluating a use case where users ask questions based on a screenshot or uploaded document, and want to understand whether multimodal support is on the roadmap.
If not, can you please recommend a specific architectural pattern to integrate this capability cleanly with vanna?
Any guidance on the recommended approach would be helpful.
Also our team will be happy to develop and contribute if anyone from vanna can guide.