The system you've been running quietly for years

There is a question I ask early in every coaching call. It is annoyingly hard to answer.

Walk me through how you actually do your work. Not what you’d say at a conference. The part nobody sees.

The expert pauses. Tries. Stops. Tries again. The first attempt always sounds like the LinkedIn version — clean, framed, generic. Recognisable, and not specifically theirs. They hear it themselves and frown.

Then I ask them to walk me through a single recent moment instead. One client meeting. One brief that landed on their desk last Tuesday. One decision they made between 3pm and 3.04pm. And slowly — sometimes uncomfortably — the actual system comes out.

That gap between what they say they do and what they actually do is where the work of this essay lives.

The visibility paradox

The more you have internalised something, the less visible it is to you. This is the cost of mastery. By the time you can do something fluently, you can no longer narrate it — not because you have stopped doing it, but because it has stopped registering as a decision. It registers as obvious.

The danger of obvious is that it makes you assume what you do is what anyone with your title would do. That assumption is wrong, and it is expensive.

Two senior consultants. Identical CVs. Same ten years at the same firm. Same client industry. They will read the same brief in wildly different ways. One will fixate on a sentence buried at the bottom of page three. The other will skim past it and zero in on the org chart instead. Their respective blind spots are visible to anyone who has watched both of them work — but invisible, often, to themselves. That second-order invisibility is the moat. It is also the problem.

If you cannot see the system you have been running, you cannot externalise it. If you cannot externalise it, AI cannot amplify it. You will outsource generic versions of yourself to the machine and wonder why the output sounds like everyone else.

Why introspection alone won’t do

The instinct, when someone asks you to map your work, is to introspect. Sit quietly. Try to remember. Distil the principles. Write the framework.

This produces LinkedIn content. It does not produce a moat.

The reason is mechanical. Introspection runs through the same cognitive process that produces post-hoc rationalisation: your brain wants a clean narrative, so it builds one. It smooths over the contradictions, filters out the embarrassing or contingent decisions, and hands you a tidy story that sounds principled but doesn’t match the way you actually work.

The work of mapping trained instinct is closer to anthropology than to introspection. You are watching yourself in the field. You are catching real moments and writing them down before your brain has a chance to neaten them. This is the discipline at the heart of phase one of The Anchor Method: Take Bearings.

Where to look: five places trained instinct hides

Once you stop trying to think your way to your own system and start watching your way there, the same five places keep showing up. Each one is a place where your trained instinct is firing — usually below the threshold of conscious narration.

One: the decisions you make in the first five minutes. What do you notice in a brief that someone three years junior would have skimmed past? What part of a prospect’s email do you re-read three times while they think you’re composing a reply? What are you ignoring — deliberately — that someone else would have flagged? The first five minutes are mostly trained instinct. They are also the easiest minutes to capture, because you can write down what happened in them right after.

Two: the moments you flinch. When something a client says triggers an internal no you can’t fully articulate — that flinch is your trained instinct firing. It is detecting a pattern that has gone wrong before. The value isn’t the flinch itself; it is the pattern behind it. Catch the flinch in the moment, then sit down at the end of the day and ask: what was the flinch about?

Three: the questions you ask that nobody else asks. Your specific questions are the externalisation of your specific framing. If you and a peer with the same job both interview the same prospect, you will not ask the same questions. The difference is your moat in motion. Note your three or four recurring questions; they are signposts to the shape of your judgement.

Four: the work you redo that nobody asked you to redo. The standards no one explicitly set, but you have imposed on yourself. The deliverables you have refused to send because they were not yet ready — even though the client would not have noticed. The phrase you keep cutting because it bothers you. The standard is the moat. Catalogue what you redo, and you will see the shape of it.

Five: the things you have quietly stopped doing. Every senior professional has a list of I used to do this and stopped. The list is gold. The reasons behind each item are gold. Your trained instinct has been pruning your own behaviour for years on signals you may never have written down. Surface the list and the reasons surface with it.

How do you capture without flattening it?

Three rules, in order of importance.

Capture in the moment, not in retrospect. The cleaner version of an event — the one your brain will volunteer at the end of the week — is rationalised and generic. The version captured between meetings, on the way out of a call, in voice notes from the carpark, is rough and specific. The rough version is what you want. Polish kills the information density.

Specificity beats generality — always. “The prospect’s CTO mentioned cost three times in casual phrasing within the first eight minutes” is useful. “I read the room well” is not. The first is a data point you can encode into a system. The second is wallpaper.

Resist three habits that flatten the data. Post-hoc rationalising. Reaching for stock industry phrases. Comparing yourself to peers. Each one tries to slot what you actually did into a category that already exists, and in doing so, erases the part that is yours specifically. Note the urge, and write what happened anyway.

For most people, two weeks of capture — a few minutes a day, after meetings — is enough to have a usable corpus. Not a long, polished document. A messy text file or notebook with maybe forty entries, each one a moment, each one specific. That is the artefact.

What the artefact is for

The artefact is not for someone else to read. It is the raw material out of which you design a system — including, at the right time, an AI system.

This is where the leverage essay connects. There you are now. Five components: a model, a context, a memory, tools, and judgement. The corpus you have just assembled is the highest-leverage thing you can put into the context and judgement components — the two parts only you can populate. A model with no trained-instinct corpus is a generic assistant. A model with one is a working externalisation of how you read the room.

This is the move that makes AI agents stop sounding like knock-offs and start sounding like you.

Take Bearings is not glamorous. It produces no posts. It looks, from the outside, like nothing is happening. From the inside, it is the most important three weeks of the path. It is where the moat becomes visible — first to you, then to the system you will build around it.

What to expect from the work

Three observations from watching domain experts go through this.

The first week feels uncomfortable. Most senior professionals have never written down what they actually do, and the experience of watching their own judgement up close is faintly embarrassing. They think they should have been able to articulate this years ago. They could not have. The thing being articulated did not exist as language until they wrote it down.

The second week, patterns emerge. The flinches stop feeling random. The questions you keep asking start to look like a small set of moves. The standards you impose on yourself start to read as a coherent set of values. You begin to see the shape.

By the third week, you have something rare and useful: a description of your work that no one else could write, in language no one else would use. From here, the next phase of the method — Charting Course — is where you turn it into a working AI system around your judgement. If you want to walk the path inside a small group with structure, the cohort is built around this exact arc.

What I want you to leave with is this:

You have been running a system for years. You have not seen it, because the system runs faster than introspection can. The way to see it is not to think harder but to watch yourself in real moments — the first five minutes of a brief, the flinches, the questions, the work you redo, the things you have quietly stopped.

Capture those, specifically. Resist the urge to tidy. Two weeks of rough notes are worth more than two months of polished framework, because the rough notes contain the information that the polish would have removed.

That set of notes is the moat made visible. Once it is visible, AI is the leverage that takes it and makes it bigger than you alone could.

Take bearings first. The rest of the path opens from there.