Level Up • Weekly Newsletter
The Mirror Principle
How the most effective leaders are using AI to see around corners — before decisions become commitments.
By Divyesh Khetia
Idea in Brief
Most consequential leadership mistakes are not failures of intelligence. They are failures of angle. The risk was there. It simply lived around a corner the leader had not yet turned.
AI, used intentionally before decisions rather than after them, can expand that angle of view by surfacing assumptions and revealing what a well-informed skeptic would see.
Red teams, pre-mortems, and adversarial advisors have existed for decades. What AI changes is how accessible and immediate that challenge becomes.
Leaders who build this as a consistent practice are developing a compounding capacity to see further than experience alone would allow.
Opening
I want to share something I have been working through personally, because I suspect some of you are navigating a version of the same challenge.
We spend a lot of time in leadership conversations talking about what we know. The data we have gathered. The experience we have accumulated. The track record we have built.
What we talk about less is what we cannot yet see. The assumption sitting just out of view. The risk that does not register because it lives around the corner from where we are currently looking.
The most valuable thing AI has done for my thinking is not help me move faster. It has helped me see further. To turn the corner on a decision before I commit to it, rather than after.
What I Have Noticed in Myself
Starting with execution. Arriving at something more.
When I first started using AI regularly, I used it the way most people do. I asked it questions I already had partial answers to. I gave it tasks I wanted completed. I used it to move faster through things I already knew how to do.
It was useful. But I was not getting the most important thing it could offer.
The shift happened when I started using it before decisions rather than after them. Not to confirm the direction I was heading but to genuinely challenge it. I would describe the decision I was facing, share the reasoning I had developed, and then ask: what is the strongest case against this? What am I not seeing? What would someone who fundamentally disagreed with this direction say, and why might they be right?
The quality of thinking that came back was not always perfect. But it was consistently useful. And more importantly, it showed me angles that are sometimes hard to find inside an organization where people are naturally inclined to build on the leader's existing direction rather than challenge it.
The Challenge Every Seasoned Leader Faces
Experience is your greatest asset. It can also narrow your frame.
Here is something I think we do not say enough in leadership conversations: experience is double-edged.
The pattern recognition that comes from years of navigating complex situations is genuinely valuable. It is what allows leaders to move with confidence in uncertain environments. It is what separates good judgment from uninformed opinion.
But that same depth of experience can narrow the frame. When you have seen a situation before, you can be tempted to reach for the playbook that worked last time without fully examining whether the conditions are the same.
The biases that affect the most senior leaders are often not the obvious ones. They are the invisible ones — assumptions so embedded in how we think that they do not register as assumptions at all.
Behavioral economics research, most notably the foundational work of the late Daniel Kahneman, whose Nobel Prize-winning insights on cognitive bias McKinsey consultants directly applied to organizational decision making, documented this pattern with precision. This is not a character flaw. It is a structural feature of how experienced minds work. And it is precisely why having a structured external challenge before major decisions matters so much.
What AI Makes Possible
A synthetic sounding board. Not a replacement for judgment — an expansion of it.
For most of leadership history, the tools for this kind of structured challenge were slow, expensive, or both. Bringing in a board subcommittee to pressure-test a strategic assumption takes weeks. Hiring a second advisory firm to argue the opposing case, the approach Warren Buffett famously employed throughout his tenure at Berkshire Hathaway, routinely commissioning one adviser to make the case for a deal and another to argue against it, is not available to every leader or every decision.
AI does not fully replace any of those things. The depth of a truly seasoned board member, the relational trust of a long-term advisor, the institutional memory of a colleague who was in the room when the last version of this decision was made — none of that is replicated by a language model.
But AI does make one thing far more accessible: the structured challenge that surfaces what you might be missing before you commit. You can run the same decision through multiple lenses — a risk-focused perspective, a competitor's view, a skeptic on the board — and get a richer picture of the decision surface in a fraction of the time those conversations would take to arrange.
The Practice, In Plain Terms
Three steps. One honest question.
Before a significant decision, write down the core assumption that would have to be true for your preferred direction to succeed. The single most important thing you are betting on.
Step TwoAsk your AI tool to make the most rigorous, well-reasoned case for why that assumption is wrong. Not a balanced view. Not pros and cons. The strongest possible challenge.
Step ThreeRead it as if it came from someone you deeply respect who fundamentally disagrees with you. Then ask: does my conviction survive this?
That is the practice. It is not complicated. It is disciplined. And it is harder than it sounds, because sitting with a well-constructed challenge to something you believe requires genuine intellectual courage.
A Word on What This Is Not
This is not about doubt. It is about durability.
This approach is not about creating doubt for its own sake. Leaders who have built strong track records have done so, in part, because they are able to move with conviction. That capacity is worth protecting.
The goal is not to undermine conviction. It is to ensure that conviction is well-founded rather than simply unchallenged.
Conviction that survives a genuine challenge is more durable than conviction that was never tested. The corner you cannot see around is usually where the risk lives.
Frequently Asked Questions
I already have strong advisors I trust. What does AI add?
Your advisors see from inside the same organizational context you do. They have relationships, stakes, and institutional histories that shape what they raise and how they raise it. That is not a flaw in them. It is the nature of the environment. AI has no stake in the outcome. That structural neutrality is what allows it to show you the angle your most trusted people might not think to show you — or might not feel safe showing you.
How is this different from doing more analysis?
More analysis adds information to an existing reasoning process. This practice examines the reasoning process itself. The goal is not to know more before deciding. It is to understand what assumptions your current knowledge is resting on — and whether those assumptions are as solid as they feel.
What if the AI's challenge is not fully accurate?
It does not need to be perfectly accurate to be useful. A challenge that is directionally right, even if imperfect in its details, will surface questions worth engaging with. The quality of your response to the challenge matters more than the perfection of the challenge itself.
Does this apply to team decisions, not just individual ones?
It is arguably more valuable in team settings. Group decisions are particularly vulnerable to positional influence, where the most senior voice in the room shapes the conclusion before genuine deliberation has occurred. Running a structured AI challenge before a team decision is brought to the table can reset the quality of that deliberation significantly.
Where do I start if I want to build this as a consistent practice?
Pick one recurring decision type where your assumptions tend to go unexamined. For some leaders it is resource allocation. For others it is talent decisions or strategic partnerships. Start with one decision in that category. Before you commit, write down your core assumption and ask for the strongest challenge to it. Do that five times. The pattern in your own thinking will become clear — and so will the value of the practice.
"Turn the corner before the decision does it for you."
If something in this shifted how you are thinking about an upcoming decision, try the practice once before you commit. Then share this with one person in your network who is facing a decision they have not fully pressure-tested. That conversation will compound.
Subscribe to Level UpLevel Up • Weekly insights for leaders who want to think further.
Published every week by Div Khetia.
