AI's biggest leverage isn't doing the same things faster. It's doing things that weren't possible before.
The dominant AI conversation has been almost exclusively about productivity. Faster reports. Faster code. Faster customer service. The problem with productivity-only framing is that it caps the upside at "the same business, but cheaper." Berkeley Haas spent eight months observing knowledge workers using AI and reached a finding that should sit uncomfortably with every transformation roadmap: people who use AI tools don't end up working less — they end up working the same amount, on more things, with the same exhaustion.
The real leverage is somewhere else. It's in the products that didn't previously exist, the workflows that weren't economically viable, the customer experiences that weren't technically feasible. We chase the non-obvious applications because that's where the actual transformation lives. Cost-cutting is table stakes.