Walk into any corporate office today and you’ll find something that would have been unimaginable three years ago: near-identical AI toolkits on nearly every laptop. The same models, the same subscriptions, the same handful of tabs open through the workday. Access, the thing everyone assumed would define the AI era, has quietly become the least interesting variable in the room.
And yet the gap between professionals who are pulling real leverage from these tools and those who aren’t is not narrowing. It’s widening. The question is why and the answer, increasingly, has little to do with technology and almost everything to do with temperament.
For most of the last two years, the dominant use case for generative AI at work has been retrieval. Summarise this. Draft that. Rewrite this paragraph in a friendlier tone. It felt like productivity, and for a while it was. But the productivity was borrowed. Everyone picked up the same habits at roughly the same speed, which meant the advantage evaporated into a new baseline almost as quickly as it appeared. Faster output stopped being a differentiator the moment it became universal.
What’s replacing it is harder to see from the outside, because it doesn’t look like productivity at all. It looks like argument.
Saurabh Khopade, a consulting professional and Certified Scrum Product Owner who works closely with enterprise systems, has been observing this shift and frames it more bluntly than most. “The people getting real value from these tools have stopped asking them questions,” Saurabh says. “They’re using them to interrogate their own thinking. There’s a difference between ‘what should I do’ and ‘what am I getting wrong’ and almost all the leverage sits in the second question.”
It’s a counterintuitive reframe, because it cuts against nearly every piece of AI training material circulating in corporate learning platforms. The standard advice is to write better prompts, add more context, be more specific. Saurabh suggests the real upgrade isn’t in the prompt – it’s in the purpose. “If you go in looking for confirmation, the model will give it to you. These systems are trained to be agreeable. The professionals I see pulling ahead are the ones who actively fight that tendency. They ask the model to disagree with them, to argue the opposite case, to find the weakest link in their own logic.”
A fair objection surfaces here: isn’t this just prompt engineering with extra steps? Saurabh has heard the pushback and rejects the framing. “Prompt engineering is about getting a better answer out of the model. This is about getting a better question into your own head. The output almost doesn’t matter – what matters is what you notice about your own thinking while you’re reading it. That’s not a prompt trick. That’s a working habit.”
The pattern, once you know to look for it, shows up in unexpected places. A product manager preparing a launch is asking the model to argue why the rollout will miss its numbers before the deck ever reaches leadership. A consultant stress-testing a recommendation is running it past the model as a hostile client would read it, hunting for the objection that hasn’t been anticipated. A founder sitting with a pricing decision is asking the model to make the case for the option they’ve already rejected, just to see if the rejection still holds. In each case, the AI isn’t producing the work. It’s producing the doubt that sharpens the work.
This, he suggests, is the part most organisations should consider in their AI rollouts. “Companies are teaching people to use AI to go faster. They should be teaching them to use it to think more carefully. Speed is table stakes now. Judgment is what scales. The firms that figure this out will quietly outperform the ones still measuring hours saved.”
What makes the shift interesting is that it can’t be bought, subscribed to, or rolled out through an enterprise licence. It’s a posture, not a product. And postures, unlike tools, don’t compress overnight. A professional who spends six months learning to argue with a model builds a habit their colleagues can’t replicate by upgrading their plan next quarter.
Which may be why the AI advantage at work is starting to look less like a technology story and more like a temperament one. The professionals quietly pulling ahead aren’t the ones with the most tools or the flashiest workflows. They’re the ones who have stopped looking for answers and started looking for holes.
“Productivity was the first chapter of the AI story,” Saurabh says. “Thinking is the second. And the professionals who treat that transition as optional will spend the rest of their careers wondering when the room moved without them.”
