"I find claims of 3x to be very suspicious." That is the CPO of an $8.3 billion AI company, whose team runs one of the more sophisticated AI workflows I have seen on this podcast. Charles Zedlewski's reasoning is worth more than the skepticism. Put a dozen engineers and a PM on something and most of what happens is discovery, debate, reevaluation, and coordination. None of that gets magically better with AI. Compressing research is real. Compressing coding and testing is real. He says that…