It is twenty to ten on a Tuesday. The board pack goes out Thursday morning. The decision sitting at the centre of it is whether to close a business line that has underperformed for six straight quarters, employs forty-one people, and was built by someone she personally recruited eight years ago.
She has been circling it for a week. Tonight, she opens a window, types four paragraphs of context, and asks for a recommendation.
Ninety seconds later she has three options, a risk matrix, a stakeholder map, and a clear recommendation with the logic laid out underneath it. It is good. Genuinely good, and better structured than anything she would have written at this hour. She reshapes a few sentences, so they sound like her. She adds two details that only she would know. She sends it.
On Thursday the board accepts the recommendation. Two directors tell her it is the clearest paper they have seen from her. She thanks them and means it.
Driving home, she feels nothing. Not relief. Not the satisfaction of a hard call made well. Just a flat quiet where something should have been. She notices it for about four seconds, decides she is tired, and puts the radio on.
That drive home is what this document is about. Not the decision. The drive home.
Leadership is entering a new era. Not because artificial intelligence is replacing leaders. Because, increasingly, leaders are replacing their own thinking.
Every day, intelligent, capable leaders are outsourcing the first step of leadership: the work of wrestling with a problem, forming a view, testing assumptions, and arriving at a judgment that is unmistakably their own. One prompt at a time, voluntarily, and it feels like progress every single time.
The answers are often excellent. That is precisely what makes the risk so hard to recognise. The output is clean, the language is confident, the turnaround is fast. Nobody in the room can tell the difference, which means no correction is coming from outside. The presentation works. The recommendation is accepted. The feedback loop returns nothing but approval.
But the leader knows. Somewhere underneath the performance, in the place where honesty lives whether you visit it or not, they know. There is a flatness to it. A hollowness behind the words that only they can feel, because they did not put themselves into it. They borrowed the thinking and signed their name to it.
This is borrowed thinking: the gradual outsourcing of human judgment in favour of convenient answers.
Not because leaders lack capability, but because modern tools make it ever easier to bypass the thinking that develops wisdom, judgment and agency.
Over time, borrowed thinking begins to feel like your own. Decisions become easier to make and harder to own. Judgment gives way to convenience. Agency is surrendered one prompt at a time, until the leader remains fully accountable for decisions they never truly built.
For a long time this was something leaders could only feel, never point to. That is changing. In 2025, researchers at the MIT Media Lab put the effect under an EEG. They had people write essays with an AI assistant, with a search engine, or with nothing but their own mind, and watched what the brain did. The unaided writers showed the strongest, most connected neural activity. The AI-assisted group showed the weakest. More telling than the brain scans were two human details: the AI group felt little ownership of the work that carried their name, and many could not quote a line of what they had just produced. The researchers gave the pattern a name — cognitive debt. It is the flatness on the drive home, measured.¹
This is not a technology problem. It is a leadership problem. And it is becoming one of the defining challenges of the AI era.
¹ Nataliya Kosmyna et al., “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task,” MIT Media Lab, 2025. arxiv.org/abs/2506.08872
Your experience, your judgment, your intuition, and your context. Technology can strengthen these things. It should never replace them.
Agency is not authority. Authority is about who reports to you, what you control, whose budget you sign. Those are positions, and positions are granted and removed by other people. Agency is different. Agency means: I own my thinking, I create my results, and I am the origin of what follows.
When a leader has agency over their thoughts, everything downstream carries weight. The decision has mass to it. The accountability is real, because you cannot properly own an outcome whose reasoning you did not build. The confidence is not performed. It is earned, from the inside out, and other people feel the difference even when they could not tell you what they are feeling.
When a leader loses agency over their thinking, even gradually, then everything downstream becomes borrowed. The decision is defensible but not owned. The accountability becomes technically true and emotionally absent. The confidence becomes a costume, and costumes are heavy to wear all day.
This is what Think First is about. Not the outputs, not the tools, not even the decisions themselves, but the thinking. Who does it, and who it belongs to.
Stated plainly, in one place. The rest of this page is the working behind them.
The decisions that keep coming back are not waiting on information. You know the people, the history and what nobody is saying, and it still will not resolve.
Insight is not a fact retrieved but a story tested until the picture reorganises. It does not arrive as I did not know that. It arrives as of course.
An answer you were given and an answer you reached are not the same thing, even when the words match. Ask a machine and you get the words without the working.
We did not lose the ability to reason. We handed it over a piece at a time, and what weakens is not what we know but our patience for sitting with a problem before solving it.
Answers are becoming cheap and judgment is not. The polish is the danger: a rough story you would question, a fluent one you accept.
The machine is not in the board meeting, does not make the call to the person affected, and is not awake at two in the morning. The decision was always yours; only the thinking was outsourced.
Almost everything built with this technology aims at the answer; aimed at the question instead, it keeps the problem open and refuses the easy close. The assumption, held every time: the smartest thing in the conversation is the human.
Give yourself a few minutes alone with the problem before you reach for anything else. Then use everything — AI, coaches, data, peers — but arrive having already begun.
A practice is defined by what it asks: attention, honesty, staying with something past the point where it is comfortable. Nothing designed to save you effort will ever ask more of you.
Most of what crosses your desk should be decided quickly, and there are faster tools for that. What is left is a smaller set of moments: the ones that keep returning, the ones you take home.
A better violin does not threaten music, and two leaders using the same model end up in completely different places. LIS is the instrument we built for this belief; the belief came first, and if something better carries it one day, every line here is still true.
This idea will be misread, and the misreadings matter more than usual, because each one is a comfortable place for a leader to stop.
That is the fearful version of this idea, and it is exactly wrong. The most sophisticated thinking humans do has always happened in relationship: with a coach, a trusted peer, a partner who asks the next question. Think First leaders use all of it, constantly. The difference is the order. They arrive at that help having already formed a view, even a rough one, and noticed what they do not know — the single most useful thing anyone can bring to a conversation.
And then they reach out, which means they ask sharper questions, receive better input, and stay in the driver's seat of whatever comes back. The leader who has thought first can tell when the answer is wrong. The leader who has not, cannot. Think First does not close the door to help. It determines who walks through it first.
This one is subtle, and it is the most important door in the room. Prompt engineering is thinking's most convincing impersonator. You have learned to frame the question more precisely, give more context, specify the format and the audience. The output improves noticeably. It feels like a skill, because it is one. But look at the sequence. You still went to the machine first. The agency did not move an inch. You simply got more efficient at handing it over, and the improved quality of the output makes the handover harder to notice.
Think First has nothing to do with how good you are at talking to AI. It is entirely about what happens before you open the window.
Reading a spreadsheet, processing data, synthesising a report — these have real value, but they are not thinking, and confusing them is one of the most common failures in senior leadership. You can analyse a situation perfectly and still miss the thing that matters most, because it is almost never in the data.
Analysis tells you that a senior leader's team has had four resignations in seven months. It does not tell you that he stopped taking risks after a meeting eighteen months ago where you publicly overrode him, and that everything since has been him protecting himself. No dataset holds that. One person in the organisation can find it, and it is you, and only if you go looking.
Nobody argued that we should refuse ATMs and go back to queuing for a teller. The ATM did not replace financial judgment. It handled the transaction. What to do with the money stayed where it always was. Nobody won by refusing the internet either. But the people who won were not the ones who used it most. They were the ones who used it to think better, rather than instead.
We are standing at the same inflection point, and it is arriving faster than the last two. Most people will use AI unconsciously. They will let it handle more than the transaction. Slowly, without ever deciding to, they will let it handle the judgment. Think First is the choice not to do that. Not because AI is dangerous, but because your judgment is irreplaceable and worth protecting.
That is a choice. Almost nobody is making it consciously.
An insight is not new information. It is not a conclusion arrived at by elimination. It is not the output of a well-constructed analysis, however impressive that analysis may be. Here is what it is.
Take a photograph of a busy intersection. A market, a crossing, a street where many things are happening at once. Ask ten people what is going on in the picture and you will get ten stories. The man at the kerb is late for work. The couple outside the café are arguing. The woman with the bag is waiting for someone who is not coming. Every story is confident. Every story is fluent. And no two are the same, which is the proof that none of them is insight. They are assumptions wearing the costume of observation.
Now imagine you could step into the scene and ask. The man at the kerb is not late for anything. He comes to this corner every year on the same date, because it is where he last saw his father. The moment you hear it, the whole picture reorganises. Nothing in the photograph changed. What changed is you. Your story met the scene, and the scene won.
That is insight. Not something invented. Not something handed over. Something reached, by testing the story you were telling yourself against what is actually there. And it arrives with a tell you cannot fake. New information arrives as I did not know that. Insight arrives as of course. This is close to what the psychologist Gary Klein found studying how insight actually happens: not a fact retrieved, but a new connection made — a story tested against reality until the picture reorganises. The photograph is not a metaphor for the research. It is the research.²
Here is what that means for leadership. In the decisions that matter, the scene is yours. It is made of pieces only you hold: the hiring conversation years ago, the exchange in a corridor in March you never consciously filed, your history with the people involved, your read of what is really being asked. No dataset contains those pieces. No model was trained on them. Which means the moment the picture reorganises can only happen in you. A tool can push you to test your story. It cannot see your scene.
And now you know what the answering machines are, in this picture. They are one more story about the intersection. A better one, usually: more fluent, better structured than the story you would have told yourself. But still a story, told from outside the scene, by something that has never spoken to anyone in it. The polish is the danger. A rough story you would have questioned. A polished one, you sign.
An answering machine gives you a better story. It cannot give you the scene.
Think First is the practice of going to the scene, which only you can reach, before accepting anyone's story. Including your own.
² Gary Klein, Seeing What Others Don’t: The Remarkable Ways We Gain Insights (PublicAffairs, 2013). psychologytoday.com/us/blog/seeing-what-others-dont
Human agency is leadership's greatest advantage. The ability to think independently, make conscious choices, own decisions, and remain the true author of your leadership is becoming more valuable, not less, in a world where answers are abundant.
Borrowed thinking feels like borrowed thinking, even when nobody else can tell. Leaders who lead from it do not collapse dramatically. They hollow out slowly, over years, in a way that is easy to attribute to workload or age or the general weight of the role.
The mechanism is not mysterious. Thinking is a practice, and practices decay when they are not used. Every problem handed over before it is engaged with is a repetition that did not happen — individually, none of them matter, which is precisely the danger. There is no single moment where skipping it is obviously the wrong call. There is only the accumulation. A 2025 study from Carnegie Mellon and Microsoft Research, drawing on nearly a thousand real examples of professionals using AI at work, found that the more a person trusted the tool, the less critical thinking they brought to what it produced. The habit erodes exactly where it is most convenient to let it.³
And there is a second reason, harder than the first. The cost does not transfer. The restructure that does not deliver. The person you moved on. The number you committed to in front of a board. Every bit of that is yours, and it stays yours.
The machine is not in the board meeting. It does not make the phone call to the person affected. It is not awake at two in the morning.
The decision was always yours and always will be. Only the thinking was outsourced, which means you are carrying the full weight of a conclusion you did not build. That arrangement holds up perfectly until it meets resistance. Then you find out which kind of answer you were carrying.
Answers are becoming cheap. Judgment is not. And the better the answers get, the more sense it makes to take them, which is exactly what makes this the hardest moment in the history of the problem. The tool is often right. That is not the issue. The issue is what you are left holding when it is.
The future will not belong to the leaders who use AI the most. It will belong to the leaders who never surrender the thinking that only they can do.
³ Hao-Ping Lee et al., “The Impact of Generative AI on Critical Thinking,” Carnegie Mellon University and Microsoft Research, CHI ’25, 2025. doi.org/10.1145/3706598.3713778
A Think First leader treats their own mind as the first instrument, not the last resort.
They do not arrive at a problem and immediately reach for an answer. They go inward first, drawing on experience, instinct, context, and their read of what is really being asked. They form a view, even a rough one. Then they reach out. When they use AI, they use it as a tool, not a replacement. When they use a coach, they arrive with something already begun. When they walk into a boardroom, the thinking is theirs, and the people in the room feel that, even when they cannot name it.
You have met them. They are not the person who dominates every silence, or arrives with the most data or the most polished recommendation. They are the person who, when they speak, makes you feel something underneath the words — a depth you sense before you can explain it. You trust what they say, not because of their credentials, but because you can tell the thinking is theirs.
That depth is not intelligence, and it is not experience alone. Plenty of experienced people have none of it. It is the accumulation of genuine thinking over time. Every occasion they went inward first. Every story they tested instead of signed. Every insight they sat with long enough to own. It compounds, and people feel it without being able to name it.
There is a paradox at the centre of the Think First leader: they carry more and say less, because they do not need to prove it. Contrast that with the leader who has outsourced their thinking — they fill space, reference the data, the framework, the benchmark, while somewhere underneath they know there is nothing load-bearing there. The words are doing work the thinking should have done.
Think First leaders are recognisable by the weight others feel when they speak. They are the origin of their own conclusions.
There is a fair question sitting in the middle of all this, and it deserves a direct answer rather than a clever one. If the argument is that leaders should stop handing their thinking to AI, why does the answer arrive in the form of an AI product?
Because the technology is not the variable. What you aim it at is.
Almost everything built with this technology aims at the answer. That is the product, and the measure of quality is how good the story is and how quickly it arrives. Used deliberately, that is enormously valuable. Used by default, it removes the leader from the process without ever announcing that it is happening, because the story is good and nobody objects.
Point the intelligence at the question.
Aim it there instead and you get something structurally different. Keep the problem open. Refuse the easy close. Notice what is being avoided. Push the leader back toward their own scene. Assume, every single time, that the smartest thing in the conversation is the human, and that the entire job is to get them thinking.
That is not AI replacing thinking. That is AI protecting it.
And it is not a philosophical distinction. It is a testable one, and here is the test.
Any moment the system takes credit for what you saw, the method has failed.
That is a hard line to hold and it is deliberately hard, because it is the only claim in this document that can be checked in about four minutes of use. One kind of tool ends with you holding an answer you did not build. The other ends with you saying something out loud that you had not quite articulated before and recognising it as true the moment you hear yourself say it.
Same underlying technology. Opposite question. The tool was never the issue. What you ask it to do to you is the issue, and that choice is still sitting there, unmade.
Go back to twenty to ten on that Tuesday. Same board pack, same six quarters, same forty-one people, same person she recruited eight years ago. She types roughly the same four paragraphs.
What comes back is not a recommendation. It is a question.
Not a generic one. Something closer to: you have described six quarters of numbers and forty-one people, and you have not once described what you think should happen here. What do you already believe about this?
She answers, slightly irritated, because she is tired and she wanted the paper drafted. The next question is worse. It asks what she is protecting.
Somewhere around the ninth exchange she types a sentence she did not know she was going to type. That she decided this three weeks ago. That the numbers stopped being the question a long time ago, and what she has been circling for a week is whether she can sit across from the person she recruited and tell him herself, rather than letting it arrive through a process.
That is the real decision. It was never in the data. It was in her scene: a hiring conversation in 2018, a corridor exchange in March, a loyalty she had not wanted to look at directly. The story she had been telling herself was about numbers. The scene was about him.
The session runs about fifteen exchanges. Then she writes the board paper herself, and it takes her forty minutes, and it is hers.
On Thursday, the board still accepts it. Two directors still tell her it is the clearest paper they have seen from her. The drive home is different. Not triumphant. Solid. She made a hard call, she knows exactly why, and she is carrying something she built rather than something she borrowed.
Three weeks later a related decision arrives, and the sentence she wrote at exchange nine is still there, in her words, exactly as she typed it. She does not re-explain the history. She starts from where she got to.
That is the difference this whole document is about. It is also the reason the instrument exists.
Not a coach. Not a chatbot. A thinking partner.
This is worth saying plainly, because overselling it would undermine the whole argument.
Most of what crosses your desk should be decided quickly, and you should decide it quickly. If you need an answer in ten minutes, use a tool built for answers. They are very good, and they are faster than this.
If you have already decided and what you want is agreement, this will be an uncomfortable hour. It will keep opening the thing you came to close.
What remains is a smaller set of moments. The decisions you take home. The calls that sit with you. The problems that live at the intersection of data, relationship, history and judgment, the place where only you can go. That is the territory this is built for. Deliberately narrow. Exactly right.
Not for every decision. For the ones you take home.
You already know how to think. You have been doing it your entire career, and the question was never capability. The question is sequence, and commitment.
Every meaningful leadership challenge begins the same way. Pause. Reflect. Form a view. Own your first thought. Then use every resource available, AI, coaches, colleagues, data and experience, to strengthen it. The sequence matters.
Four minutes is enough to change the sequence. Sit with the actual problem and ask yourself:
What do I already know about this that I have not said out loud to anyone?
What story am I telling myself about this, and have I tested it against the scene?
What am I avoiding looking at directly?
Is the question I have been handed the decision in front of me, or is the real decision sitting somewhere underneath it?
Notice what you know. Notice what feels genuinely uncertain, as distinct from what feels uncomfortable. Notice the thing you keep sliding past. Then reach out and use everything available to you. But arrive having already done something. Having already put yourself into the problem.
Think First is not about rejecting help. It is about remaining the author of your own judgment.
Instruments improve, and practices absorb the improvement. A better violin does not threaten music. It goes to whoever has the practice, and it does almost nothing for anyone who does not. Two leaders using the same model end up in completely different places, and the difference is not the model.
LIS is the instrument we built for this belief. It is not the belief. The belief came first, and if something better carries it one day, every line in this document is still true.
Because the future of leadership will not be defined by the quality of artificial intelligence. It will be defined by the quality of human judgment that remains behind it.
Think First. Everything else follows.
Try LISThree findings are part of the argument in this document. On borrowed thinking and cognitive debt: Kosmyna et al., “Your Brain on ChatGPT,” MIT Media Lab, 2025. On the erosion of critical thinking with AI reliance: Lee et al., “The Impact of Generative AI on Critical Thinking,” Carnegie Mellon University and Microsoft Research, 2025 (CHI ’25). On the nature of insight: Gary Klein, Seeing What Others Don’t, 2013. The case here does not rest on these studies; they confirm, from the outside, what leaders have been describing from the inside.