- Tension: The entire decision-making toolkit — pros and cons lists, decision matrices, expected value calculations — rests on one assumption: that you already know what you’re optimising for. When a decision is significant enough to matter, that assumption is almost always wrong.
- Noise: A prompt circulating in AI productivity communities is being treated as a workflow tip: ask your model to describe what kind of person would choose each option. Most coverage stops at the technique.
- Direct Message: It’s not a technique. It’s a diagnosis. Significant decisions are character questions wearing an analytical costume. The prompt works because it stops pretending otherwise — and that distinction has consequences well beyond how you use AI.
There is a particular kind of paralysis that visits almost everyone who has ever faced a genuinely difficult decision: you have made the list, run the numbers, and consulted the matrix, and you are no closer to knowing what to do. The pros column and the cons column stare back at you with roughly equal weight. The expected value calculation yields a marginal advantage for one option that you are not sure you trust. Something about the whole exercise feels like it is measuring the wrong thing, but you cannot quite say what.
A prompt circulating recently in AI-assisted thinking communities put a name to that feeling. As Wyndo noted on Substack — writing under the newsletter “The AI Maker” — the insight arrived through a specific way of framing a decision to an AI model. The prompt was this:
“I have two options. Instead of telling me which is better, tell me what kind of person would choose each one.”
The effect, as Wyndo described it, was immediate and disorienting in the best sense: “Suddenly the decision wasn’t about pros and cons. It was about who I wanted to be. The best decisions aren’t logical. They’re identity-driven.” The observation is simple enough to dismiss as a productivity aphorism. It deserves more than that.
What the standard frameworks get wrong
The dominant vocabulary of decision-making — pros and cons lists, decision matrices, expected value calculations, second-order thinking — rests on a foundational assumption that is rarely examined: that you already know what you are optimizing for. These tools are extraordinarily useful when the objective is clear and the variables are measurable. A procurement decision, a build-versus-buy question, a marketing spend allocation: these are amenable to analytical frameworks because the goal function is defined in advance and the inputs are at least partially quantifiable.
But when the decision is significant — whether to leave a company you built, whether to take a role in a different country, whether to pursue a partnership that would reshape what your company does — the framework breaks down precisely because the objective function is what is at stake. You are not choosing between two known quantities of utility. You are choosing between two versions of who you might become. And for that problem, the pros and cons list is not merely imprecise. It is the wrong instrument entirely.
Research on constructed preferences — developed across decades by Daniel Kahneman, Amos Tversky, and Paul Slovic, among others — gives this intuition a formal grounding. Their work has demonstrated consistently that human beings do not, in general, have stable, pre-formed preferences sitting in memory that we access when a decision requires them. Instead, we construct preferences in real time — and the frame we use to approach a decision determines, to a significant degree, the preference we end up with. Ask someone whether they prefer Option A to Option B and you will get one answer. Show them Option A alongside a decoy, or change the default, or alter the sequence in which options are presented, and you may get a completely different answer — not because the person changed their mind, but because the preference did not exist in a fixed form to begin with.
This is not a quirk of experimental psychology. It describes how most people actually make decisions, most of the time. And it has a direct implication for the limits of analytical decision-making: if your preferences are constructed by the framing you use, then choosing a purely analytical frame does not give you access to some deeper, truer preference. It gives you the preference that analytical framing produces — which often doesn’t correspond to what actually matters to you.
The identity frame and why it works differently
The concept of self-signaling, developed by the economists Ronit Bodner and Drazen Prelec, offers a more precise account of what is happening when a decision feels like it is really about identity. Their argument, developed across a series of papers on behavioral economics and self-concept, is that every choice a person makes is simultaneously a choice about what that choice signals to themselves about who they are. You are not merely selecting an outcome. You are sending yourself a message about the kind of person who would make this choice — and that self-signal carries motivational weight that is often independent of, and sometimes more powerful than, the material consequences of the choice itself.
This is why people will donate to causes they believe in even when they are certain no one is watching, and why a founder might turn down an acquisition offer at a number that would make most rational observers raise their eyebrows. It is also why, in the Bodner-Prelec framework, self-deception is not always irrational: it can serve the purpose of maintaining a self-image that generates better behavior over time. The signal you send yourself matters, because you live with the self you are constructing through your choices.
The Wyndo reframe engages this mechanism directly. By asking not “which option has better expected outcomes?” but “what kind of person chooses each of these?”, it shifts the question from outcome space to identity space — and in doing so, it often produces a clarity that the analytical approach cannot. Not because the identity answer is more accurate in some objective sense, but because it accesses a different kind of knowledge: the knowledge of who you are trying to become, which is frequently the real decision that is hiding inside the apparent decision.
What this means for founders and professionals
For founders and senior operators, the practical consequences of this are significant. A great deal of strategic decision-making in organizations is framed analytically when it is, at root, a question about character — about what kind of company this is going to be. Which customer segment to serve. Whether to compete on price or on quality. Whether to move fast on a market opportunity at the cost of a relationship that took years to build. These questions can be dressed up as market analysis and financial modeling, and that work is not useless. But the analysis rarely resolves the decision, because the decision is not really about the numbers. It is about what the leadership team believes their company should stand for.
When founders describe being stuck on a decision — and many do, in the particular way that signals something deeper than simple uncertainty about facts — they are often stuck not because they lack information but because they are asking an analytical question about something that is fundamentally a character question. The data will not answer it. The framework will not answer it. What might answer it is the question Wyndo’s prompt asks: if you chose Option A, what kind of company would you be becoming? If you chose Option B?
This is also why certain decisions feel wrong even when the analysis says they are right. The analysis might be correct on its own terms, and the choice might still be a mismatch with the self or the organization you are in the process of building. That feeling of wrongness is not irrationality to be overcome. It is information — specifically, information that the analytical frame is not equipped to capture, because what it is registering is an identity inconsistency that only shows up when you look at decisions through a character lens.
The AI angle: reframing, not optimizing
The context in which Wyndo surfaced this prompt matters for a Silicon Canals audience. The instinct when using an AI model for decision support is to reach for the optimizer — to ask it to weigh the options, identify factors you may have missed, and produce a recommendation. This is not a bad use of the technology. But it imports the same assumption that limits the analytical approach: that the goal is to find the option with the highest score on some implicit objective function, and that the AI has or can infer what that function is.
The value of the identity reframe is not that the AI knows something the user does not. It is that a different question produces a different cognitive mode. When you ask an AI to compare options on their merits, you get analysis. When you ask it to characterize the kind of person who would make each choice, you get a mirror — and a mirror can show you things that a scorecard cannot. The reframe is a prompt engineering insight, but the principle it applies is much older: the question you ask determines the kind of answer you are capable of finding.
This has implications for how teams use AI in strategic contexts. The tools are increasingly capable of doing the analytical work — synthesizing data, modeling scenarios, pressure-testing assumptions. What they can also do, if prompted correctly, is help individuals and leadership teams engage with the identity questions that analytical work tends to suppress. That is a different kind of value, and it is underused.
Which decisions this applies to — and which it does not
The case being made here is not that all decisions should be made on identity grounds, or that analysis is a distraction from the real question of who you want to become. That would be a different and less defensible argument. Many decisions should be made analytically. Operational decisions, financial decisions, many hiring decisions — these benefit from rigor, and importing identity considerations into choices where they do not belong can produce its own distortions. The person who turns down a cost-saving decision because it does not feel consistent with their self-image as a bold operator is not accessing superior wisdom. They are using identity reasoning in a context where it does not apply.
The useful distinction is between decisions where the objective function is genuinely known and agreed upon in advance, and decisions where the apparent objective function is concealing a deeper question about values and direction. The former class is larger than people sometimes admit, and analytical frameworks work well there. The latter class — particularly at pivotal moments in a career or a company’s development — is where identity-based reasoning becomes not just relevant but prior to everything else.
Knowing which kind of decision you are facing is itself a skill, and one that the standard toolkit of decision-making advice does little to cultivate. The signal, more often than not, is the paralysis: the sense that you have done the analytical work and the answer is still not there. That absence is often not a failure of analysis. It is a sign that the question you are trying to answer analytically is one that analysis was never equipped to resolve.
The prompt is a place to start, not a formula for every decision. But for the decisions that matter most — the ones where who you are becoming is what is actually at stake — it may be the right instrument that most frameworks have been conspicuously missing.
