I have an image challenge and immediately thought of you. I have a Claude Cowork content flow set up which is working really nicely. Where I am battling with is getting a consistent style of image via the API rather than working directly in the platform. The image style works best inside ChatGPT so I've linked the API but just trying to prompt is not working as I can't give a reference image for it to see. I have also added the fal.ai API so have access to those models but then the style does not come through. Do you have any ideas on this? Thanks so much