Because the value is in what the wrapper helps your specific customer base do. Whether that's through prompt engineering or RAG, or some other technique, the v

Because the value is in what the wrapper helps your specific customer base do. Whether that's through prompt engineering or RAG, or some other technique, the value is in someone else spending the time to think of a UX that helps accomplish a specific job. If you know what you want and how to do it, then likely fiddling around with prompts can get you pretty close, but people are busy and don't have time to test asking chatGPT 500 different prompts to get one that works well, or hunt around for examples to feed into context or into a RAG index. There's an asymmetry of information and effort developing the pipeline/context/prompt construction involved. And this is what add's value. Not simply a chat window and passing whatever the user types in, to openAI's API.