Why AI spending isn’t translating to results

There’s a lot of talk right now about companies deploying AI tools and not seeing results that map to the spend. Many reasons get cited. But I suspect the simple one at the heart of it is constraints. The analogy comes from Eli Goldratt’s famous book “The Goal”, a pre-read in most introductory operations management classes. If you’re manufacturing a car and you figure out how to produce doors more efficiently — great. But if doors aren’t your bottleneck, it doesn’t matter. The body still needs only four doors. Making more doesn’t move the car out the door faster. AI tools work the same way. They can genuinely optimize many parts of a workflow. But if that workflow isn’t the constraint, optimizing it changes nothing at the outcome level. Product teams shipping more features is only valuable if features were the bottleneck — if those features unlock new markets or make a step-change improvement to customer value. If they don’t, you’ve just made the wrong thing faster. And that’s before…