Power Platform Community Forum Thread Details

I uploaded an image and Copilot appears unable to detect the alt text which is shown in its metadata. I was hoping to use it to batch extract images and alt text from PDFs so that they could be reviewed by experts, so this would be a roadblock. Is this a known issue?

hello, since it seem to be close from you'r previous question few minutes ago, i will post the same answer :) it's not a technical issue but a temporary design choice (it's not the only one, as sample metadata like author, date, etc are only very recently use to filter with "out of the box" agent) - could hope alt text will come in a near futur since they manage to add metadatas like author :)

here is my previous answer from you'r previous post :

hello,

At the moment "out of the box" agent don't recognize by default those metadatas, it's sad. In a pro environnement it's possible by making a custom agent with microsoft foundry, the "developper" need to read those metadatas and add them ot the "index" use by the IA.

But going this way "custom IA" will go further because custom IA could use "vision" to describe image and add the description of the image inside the index. This is what it's commonly use with pro agent "adding image description" (it work for image or technical schematics). But sadly at the moment it's not by default in many AI. Could be the case in the futur.

So if this a requierement for you'r company need they could use custom AI in foundry to include image description (and alt text if needed).

For web site, this week a new protocol has been released called "web mcp" it s supposed to automate web site scrapping and i hope it will include alt text in image in the web (if not it's very sad)

I hope this help you to understand and find a way to solve you'r case, if yes please mark this answer with the checkbox good answer, it's important for the community and search engine :)