Feature Request: Add Support for Attachments (Image, PDF, etc.) in Text Chat to Leverage GPT-4O Capabilities

[ashiquemukkil]

Enhance the existing text chat functionality to allow users to upload attachments (e.g., images, PDFs) alongside their text input. This will enable leveraging GPT-4O’s capabilities to process and analyze multimedia inputs in addition to textual data

Motivation
Currently, the chat only supports text input. Adding support for attachments will significantly expand the use cases, allowing users to:

• Extract information from images or documents.
• Perform multimodal analyses combining text and uploaded media.
• Improve the overall user experience by enabling richer and more interactive queries.

It will be an extra button alone with send, mic buttons, and attachments are uploaded into a separate blob container displayed on chat as well

[placerda]

File Upload Options

• PDF support
• How to store the file
• OCR
• Chunking
• Embedding
• How to search the content to be used in grounding (in-memory vector seach)
• Upload to the app service backend
• Upload the file to a blob container with expiration date.
• Create a temporary document in CosmosDB.
• Rely on Session to correlate to the CosmosDB document and files.
• OCR it with (Doc Intel or some OSS library)
• Save file text content in an attachments cosmosDB container.