@Kyle Nicol thanks. Yes, that’s what I’m doing. But that process is too time consuming. Open ChatGPT (using a custom gpt), copy my social media post, paste it in gpt, go back and forth with it, generate image. Takes too long and breaks my “flow” if you can call it that. I already have a workflow with an agent that goes through my own content (via RAG), and some other guidelines I’ve prompted with, to generate posts and put them into airtable. It generates 3-5 posts this way every day for me. I choose one and edit it lightly inside airtable. Then I’d like to fire another workflow which does the same for images. But I just can’t get the agent to generate images that are even remotely passable and that actually go well with the post I gave it. If I could do this part, I’d have a fully automated workflow that I only interact with through airtable and for which I can batch edit 20 posts in one hour, and then batch review all the images in another 10 minutes. All without leaving airtable. I know, first world problems but curious if anyone had a working prompt.

@Kyle Nicol thanks man! I’ll definitely take this for a spin and see if it works. It’s definitely the prompt agent that’s not grasping the details so I’m hopeful this will get me closer to it. I’ll let you know how it goes in the morning.

Looking forward to this build