The Palantir AIP Logic documentation doesnt have clear details on how to use Vision based models to send images and prompt with questions to get answers about the image.
this solution worked for me https://community.palantir.com/t/how-to-add-image-as-input-type-to-aip-logic-function/3421 , but I think its little scrappy and I hope the Platform team can build a better solution for users to select an object and select the media Property and send the media property as a input in the LLM input text and AIP logic will convert the image to base64 and send to LLM and get the response.