The resulting prompt is pretty accurate. I have tried with two versions of the same photo, one with its original background and one without it (transparent) and I received two different meaningful prompts accordingly. The pricing is a bit off, in my opinion: it makes the top tier the best option but that might discourage the occasional user because they might never be using 500 prompts while the entry tier at 20 cents per prompt seems a bit overpriced in comparison. The "issue" with AI images is that the generated photos are still far away from usable in a professional environment and several tries and fine-tuning are required before getting a decent result; this app certainly create good prompts and it has room for improvements but ultimately its value will increase in time when AI generated images will also get better. To me in order to keep that pricing model additional features need to be added. For example, if it generates the prompt, why not offering also the change to use it on site to generate the next photo? Or why not transforming it into an API that can be integrated with existing AI image generators? Just my idea, of course.
One thing I'd definitely consider adding though is suggesting aperture, focus and other tech camera settings, when user upload a real photo. I don't know how the system works in the background, hence if this is easily implementable or not, but in order to get a realistic photo back from an AI generator, those parameters and especially the correct ones are paramount.
All in all, this has potential and it is well crafted so considering that this is a first launch, I definitely support it.