i once worked for a small startup that had licensed an autocad clone [https://www.intellicad.org/] and were customising it as a low-cost autocad alternative for the indian architect market. the platform itself had a bunch of bugs and performance issues, and i was hired to work on some of the more egregious ones (it was kind of fun, i learnt a lot doing it, but it was an ultimately futile task because we were never going to really compete with autocad. but that wasn't the useless part; i was willing to give that my best shot as long as everyone was realistic about what we could and couldn't achieve).

the useless part came about when the company was clearly going down the drain, and the ceo was desperate to find a pivot. he had a friend who was a chip engineer, and said friend had complained to him about how hard and unintuitive EDA tools were to work with. our man gets convinced that the problem is that "these EDA firms hire scientists and mathematicians, but no one who really knows about UI design, unlike CAD companies", and that if we could hook our CAD product up to some open source circuit simulation libraries we could come up with an mvp that would be clearly so much better to use that we could get funded to develop it.

now i usually take the stance of "okay, you're the product guy, tell me what to implement and i'll get it done", but this time i pushed back pretty hard over what a fool's errand this was. but he had his hardware engineer buddy talking into one ear and the "make money fast" shoulder devil talking into the other, and would not listen. so fine, i took a couple of months to play around with CAD fileformat parsers and electronic simulation libraries, got some very basic circuits working and handed over an MVP which of course went precisely nowhere.