A headline landed in my inbox this weekend that I have been waiting two years to see. Published April 19 in CEOWORLD Magazine: “Leaders Can No Longer Fake an AI Strategy.” Microsoft published something almost identical this month: “Why AI Is an Operating Model Shift, Not a Technology Upgrade.”
I read both pieces and felt two things simultaneously. Relief that the conversation is finally catching up to what I have been seeing from the inside for years. And urgency, because the gap between knowing you can no longer fake it and knowing what real actually looks like is exactly where most organizations are going to get stuck next.
Let me try to close that gap.
For the past two years, I have watched some of the most sophisticated enterprises in the world, firms I have spent 25 years advising inside Goldman Sachs, Fidelity, American Express, and Mastercard, run what I can only describe as AI theater. They announced AI investments. They stood up governance committees. They rolled out Copilot licenses. They measured adoption rates and reported to boards that the transformation was underway.
Eighty-two percent of enterprise leaders now use generative AI at least weekly. That number has been cited everywhere this month as evidence of progress. I want to offer a different interpretation: widespread usage of a tool is not evidence of transformation. It is evidence that the tool is available. The two are not the same, and confusing them is how you end up spending two years on AI theater while your competitors are redesigning their operating models.
The era of fake AI strategy did not end because executives stopped pretending. It ended because the pretending stopped producing results that satisfied boards and CFOs who are now asking harder questions. The novelty justification, “we are learning and experimenting,” has a shelf life, and it has expired.
So what does real look like?
Real looks like starting with the replacement question, not the capability question.
The organizations I have seen get this right in the past eighteen months did not begin by asking what AI can do. They asked what specifically, inside their existing operation, should no longer require a human. That is a completely different frame, and it leads to a completely different kind of work.
When you start with replacement, you map your workflows not as they currently function but as they would function if you were designing them from scratch today. You look at your decision flows, your information pathways, the moments where intelligence is being applied to problems that do not actually require judgment, only processing. You find what I have come to call the information processing masquerading as knowledge work: the compliance reports that six people contribute to over three days, the client briefings that pass through four layers before anyone sees them, the weekly summaries that travel up three management levels before reaching the executive who needs them.
None of that is knowledge work. It is information movement wearing a title.
When you name it clearly, you find the rebuild opportunity. Not the automation opportunity. The rebuild opportunity. The chance to design a workflow that does not have those steps in it at all, because AI handles what those steps were doing and your people do what AI cannot.
Three tests I use to tell whether an AI strategy is real or theater.
The first test: can you name what AI replaced, not just what AI was added to? If the answer is a list of tools deployed and adoption rates achieved, it is theater. If the answer is a list of workflows that no longer exist in their prior form and a description of what those employees are doing instead, it is real.
The second test: has the headcount or the workflow structure changed, or did AI arrive alongside everything that already existed? Real AI transformation changes what people do, not just what tools sit next to what people already do. If the operating model looks the same as it did eighteen months ago, AI has been a decoration.
The third test: can the CFO point to where the AI investment shows up in the unit economics? Not in productivity surveys. Not in employee satisfaction scores. In the actual numbers. Real transformation produces structural cost improvement or structural capability improvement that competitors cannot easily replicate. If neither is visible yet, the strategy is still theoretical.
I want to be honest about something, because I think it is important for the conversation that is now starting.
The shift from AI theater to real AI-native transformation is not primarily a technology challenge. The technology is ready. The challenge is organizational. It requires leaders who are willing to look at their operations honestly, name what should not exist, and redesign around what does. That work is uncomfortable because it reveals that a meaningful portion of what passes for strategic work in most organizations is really information processing, and that admission has implications for teams, structures, and the stories organizations have been telling themselves about the nature of their own work.
The leaders who do this work now will have a structural advantage that compounds. The ones who upgrade their language from “AI adoption” to “AI transformation” without changing the underlying work will have spent another year on better theater.
The era of fake AI strategy is over. The era of actually doing the hard work has started.
The question is which era your organization is entering.
Keep Growing. Gunjan
