LOL - that's exactly the use case I was thinking of. But wouldn't an AI have to be trained with lots of examples of the item in question to get a high quality detection? If so, it might not be able to find that one image that antique shop in Whereverston has on their web site.

I don't work on the image side but from what I understand the entire index is vectorized so its not categorizing them like a imagenet system would as much as finding a nearest neighbor that can be categorized.