Can you provide details on wich dataset did you used for your AI? I am curious to know... why you have created the same tool for many crypto and not just focusing on some of the majors?  Likewise some of these coins can have some "insider trading" or other activities hard to be predicted by AI... Others can have very low volume or significant spread between exchanges.
How you have trained your AI for these situations? Did you plan to release for free or at least to publish this material at a certain point?