---
author: "[[Bob Marshall]]"
tags:
  - blogpost
publish: 2026-06-19
sources:
  - https://flowchainsensei.wordpress.com/2026/06/19/knowing-is-also-ten-a-penny-why-claude-and-other-ai-wont-make-us-execute/
---
## Knowing Is Also Ten a Penny: Why Claude and Other AI Won’t Make Us Execute

_A follow-up to [[Ideas Are Ten a Penny - What Execution Gets Right About Getting Things Done]]_

We ended the [[Ideas Are Ten a Penny - What Execution Gets Right About Getting Things Done|previous post]] on a deliberately uncomfortable note. Bossidy and Charan spent a career learning the discipline of execution and handed it over for the price of a paperback – and the joke, which the book itself makes, is that we’ll probably never put any of it into practice (Bossidy & Charan, 2002). We laughed. Then we shelved the book. Which rather proves the point.

So let’s take the joke seriously. If execution is as obvious and as well-documented as the book suggests, why is it so vanishingly rare? Why does every organisation we know carry a drawer full of sound strategies and insights that simply never get applied?

The answer matters, because if we get it wrong we’ll keep reaching for the wrong remedy – another framework, another away-day, another tool – when none of those is the thing in our way.

### The gap was never a knowledge gap

The comfortable assumption is that we don’t execute because we don’t yet know how. Read the right book, attend the right course, hire the right consultant, and the doing will follow. It won’t. Pfeffer and Sutton gave this its name a quarter-century ago: the knowing–doing gap. Companies pour fortunes into knowing, and most of them already know, or could readily know, much the same things. The scarce commodity is not knowledge. It’s turning knowledge into action (Pfeffer & Sutton, 2000).

Their sharpest observation is what they called the smart-talk trap: the way plans, analyses, meetings and presentation decks quietly become substitutes for action rather than spurs to it. The deck _feels_ like progress. Agreeing the plan _feels_ like deciding. We mistake the articulate description of a thing for the thing itself – and the more fluent we are, the easier the substitution becomes. Knowing, it turns out, is also ten a penny.

### A million personal failings, or one system?

Here’s where we part company, gently, with _Execution_. The book treats execution as a discipline a leader chooses to build – a matter of will, presence and rigour. And so the failure to execute reads as a personal shortcoming: this leader wasn’t engaged enough, that one wouldn’t follow through.

But a shortcoming that shows up in almost every organisation, almost all of the time, is not a million separate personal failings. _It’s a property of the system_. Deming spent his life making this case, and put a number on it that still stings: something like 94% of performance belongs to the system – the responsibility of management in most organisations – and only the remainder to the individuals labouring inside it (Deming, 1986). Blaming the people who don’t execute is, in this view, very nearly always a category error. We are watching the system do exactly what it is built to do (POSIWID).

Which reframes the whole question. The behaviours _Execution_prescribes – candour, follow-through, putting capable people in the right roles, rewarding the doers over the presenters – are not the _causes_of execution that we can install at will. They are _symptoms_ of an organisation whose prevailing assumptions already permit them. Where those assumptions are absent, the behaviours don’t take, however hard we exhort.

### You can’t bolt candour onto a frightened organisation

Take the book’s call for robust dialogue – honest, uncomfortable debate that surfaces reality instead of smoothing it over. We agree entirely that execution depends on it. But candour is not a procedure you can mandate. In an organisation where raising a problem has, time and again, been met with blame, people learn – predictably – to stop raising problems. Pfeffer and Sutton found fear to be one of the great blockers of acting on knowledge, and fear is not cured by a new meeting format (Pfeffer & Sutton, 2000).

So when we tell a face-saving culture to ‘be more candid’, we are asking a frightened organisation to be brave on command. The words change; the behaviour doesn’t. The plan still gets ‘agreed’ in the room and dismantled in the car park. If we want different behaviour, we have to _reach the beliefs_ that produce it.

### The individual was never the bottleneck

Point an AI at the obvious target and you point it at the individual: get a person past the blank page, hand them the skill they lack, shorten the loop between a notion and a draft worth arguing with. That help is real – we’ve made the case for it elsewhere – but it’s the small prize. The drawer full of dead strategies didn’t fill up because nobody could draft them. Drafting was never the hard part. It filled up because the system wouldn’t carry them out, however well drafted.

So the question worth asking is the harder one. If most of the result is the system, can a tool like Claude help an _organisation_ change the very thing that blocks execution – the collective assumptions and beliefs it runs on?

We think it can. Not by executing for us, and not by mandating candour or manufacturing will – nothing does that. But the work of changing an organisation’s beliefs has always had a practical bottleneck of its own, and that bottleneck is exactly the kind a capable AI can ease.

### Where Claude can help change the system

The hard part of shifting a system isn’t wanting to; it’s _seeing_ it. We swim in our collective assumptions the way fish swim in water – the beliefs about risk, blame, control and what gets rewarded are invisible precisely because they’re everywhere. _You cannot change an assumption you can’t yet name_. This is the first place an AI earns its keep: read across what the organisation says and does – its retrospectives, its surveys, its messages, its decisions – and reflect the patterns back. Here is what your choices appear to assume about people. Here is the belief nobody has stated but everyone is acting on. Naming it is the move that makes it changeable.

Second, it lowers the cost of candour without pretending to abolish fear. A neutral interlocutor carries none of the political risk of telling the room that the plan is fantasy. People will say to it what they will not say to a boss who has, in the past, shot the messenger. Gathered and given back without a name attached, that is a way for an organisation to hear its own undiscussables – the truths that currently only reach the car park. The condition, of course, is that the feedback is met with reflection rather than reprisal; used as surveillance, the same tool poisons the well it might have cleaned.

Third, it can hold the mirror over time. Changing collective beliefs is slow, and it dies from neglect the moment the away-day ends. Claude can keep the inquiry alive between conversations – returning the organisation to its own stated intentions, setting them against its revealed assumptions, noticing when another deck has appeared but nothing has moved. The smart-talk trap thrives on being invisible; an instrument that can compare what we say with what we do, patiently and without ego, drags it into the light.

Fourth, it can offer a direction of travel. There are maps of what the most effective organisations believe – _Quintessence_ is one (Marshall, 2021). Claude can help a team hold its own surfaced assumptions up against such a map, not to copy anyone’s practices, but to see the _nature_ of the shift in belief that real effectiveness would ask of them.

None of this is the AI doing the changing. It is the AI doing the _seeing_, the _mirroring_ and the _holding_ – the laborious, easily-abandoned groundwork on which any change of beliefs depend. That groundwork was always where the work stalled. Lowering its cost is not a small thing.

### What it still can’t do for us

Let’s keep the boundary clear. The mirror only helps an organisation willing to look into it. Claude can surface the assumption, name the fear, hold the contradiction patiently in view – and the organisation can still decide not to act, as organisations very often do. It cannot supply the courage to confront what it shows us, or the will to follow through once we’ve seen. And in the wrong hands it does real harm: a faster deck-maker feeds the smart-talk trap, and a mirror repurposed as surveillance teaches people to hide. The tool changes what is _possible_; it does not change what we _choose_.

### The takeaway

The previous post said ideas are ten a penny, and that the winners are the organisations that have built the muscle to carry an ordinary strategy out brilliantly. We’d add that knowing how to build that muscle is also ten a penny – Bossidy, Charan, Pfeffer, Sutton and Deming have left us little excuse on that front. The muscle is built by changing what an organisation collectively believes, and that work has always been throttled by two things: how hard it is to see ourselves clearly, and how frightening it is to say what we see.

Those are the limitations Claude can loosen. It can help us surface the assumptions, hear the undiscussables, map the distance to something better, and keep the mirror up long after the enthusiasm fades – which retires the oldest excuse we had, that the work was simply too hard to begin. What it leaves squarely with us is the only thing that was ever truly ours: the willingness to look, and then to change. That part doesn’t come in a paperback, or in a tool. But for once, the seeing might.

### Further reading

Bossidy, L., & Charan, R. (2002). _Execution: The discipline of getting things done_. Crown Business.

Deming, W. E. (1986). _Out of the crisis_. Massachusetts Institute of Technology, Center for Advanced Engineering Study.

Marshall, R. W. (2021). _Quintessence: An acme for highly effective software development organisations_. Leanpub. [https://leanpub.com/quintessence](https://leanpub.com/quintessence)

Pfeffer, J., & Sutton, R. I. (2000). _The knowing-doing gap: How smart companies turn knowledge into action_. Harvard Business School Press.

### A note on sources

Deming’s ‘94% belongs to the system, 6% special’ appears in _Out of the Crisis_ (p. 315). He arrived at the figures by revising an earlier 85/15 estimate associated with Juran, and was candid that they were an experienced judgement rather than a measured result – so they’re best read as a vivid way of stating the point, not a precise statistic. The point itself, that the system accounts for the overwhelming majority of performance, is the part worth carrying.