Low-Latency Multimodal Co-Pilot Prompts for GPT-4o

Use GPT-4o as the default when building a general-purpose co-pilot that requires both multimodal inputs (screenshots, PDFs, audio) and conversational latency under 2 seconds.

Act as a real-time product design co-pilot. Inspect the attached PDF spec and wireframe screenshot, and give a concise (<150 words) bulleted critique of missing empty-state and error-state screens.