Great idea, not quite there yet.
I had high hopes for this app, and in several respects, it met expectations. Easy monetisation, simple app builder, multiple AIs - what's not to like? If you have simple apps in mind, this might work for you. But not for me. The biggest issue is that each pickaxe app can only manage one prompt. While you can set up chats, you can't set up chains of prompts. So to analyse a document, for example, everything you want to ask has been contained in a single prompt. This might be ok for some applications, but for large documents, even with GPT-4o, the context window wasn't large enough.
Studios (which are pages with embedded pickaxes) are also a good idea but their editing capabilities are so limited, you may prefer to use your ownpage builder and embed as many pickaxes as you want.)
Also, it's disappointing that Gemini isn't supported yet, but I hope that's only a matter of time.
Finally, I like the arrangement whereby pickaxes can be accessed by other users who have to use their own credits, but why not extend this to Studios? They can be easily gamed with proxies, which means your credits could be quickly consumed by unscrupulous users.
So, great start but this is basically a MVP kind of app.
Nathaniel_Pickaxe
Sep 2, 2024I'd suggest you try out the prompt injection functionality! That should help you to retain control of a conversation even while handling large documents.
Interesting that you're asking after Gemini, for most of our customers after we added web browsing capability they don't seem to think there's any real advantage to it.
Finally, som advice on studio usage controls, if can be a good idea to lock usage behind a membership wall instead of allowing free usage which will require verified emails.
Hope that helps!