Markdown image syntax is notoriously hard. With multimodal input in the Prompt API, your browser can ""see"" images locally to return a structured JSON object with alt text and captions. No server latency, just private, reliable data to build your <figure> tags → https://goo.gle/4uMFh4e