I’ve attached images in multiple prompts asking the AI to follow the design, but it keeps replying that it can’t see them—even though the uploads were successful. This breaks visual-based workflows and needs to be fixed so the AI can reliably access and respond to attachments.
