How many times have any of us created a PoC that gets deployed into production? Things fall down, and are attempted to be improved like changing a tire on car driving 60mph down the highway without pulling over.
Do you stop the car, take a look at everything, make changes, and then put the car back on the road? Do you pull into pit row, change the tires in <10s with top off of the tank, and then get back in the race? Do you take advantage of the run flat tires you chose in advance anticipating the this would be the likely outcome while hoping the other 3 tires stay inflated?
If the current situation isn't work, the worst thing to do is nothing. Paralysis by analysis can be deadly
If your idea only succeeds when the smartest people are running it, and they can’t make any mistakes, then it’s not actually a good idea.