Designing Async Workflows for Multi-Model AI Image Generation

AI image generation is easy to prototype as a single API call. A user writes a prompt, the server calls a model provider, and the frontend waits for a result. That shape is fine for a demo. It becomes