My favourite case of an MVP was with a startup doing video analytics for retail. When they came to me (back in 2014), they had the network trained but no product that could have been used by the end user.
The founder said:
"I need 6 months to finish the product, then I'll go and sell it."
I pushed her out to a store near her home. She went there, copied video from security cameras, and sent it to her developers. They ran the net over this video and sent her back a few numbers.
She created a fancy report in MS Word using these numbers, printed it out on 2 sheets of paper, went back to the store, and sold these 2 sheets of paper for $1,000.
Repeated it a few times more, thus validating her hypothesis.