Closing the Knowing-Doing Gap

Twenty years ago, an idea that stayed in your head was worth nothing. I learned that the hard way, and it took me most of a career to understand why.

For as long as I can remember, I have generated ideas faster than I could act on them. The spark arrives fully formed, vivid, obviously useful, and then it is gone. Months later I watch someone else ship the exact thing I thought of first. Not because they were smarter. Because they executed, and I did not. If you are the kind of person who has lived this, you know the particular sting of it: the problem was never that you could not see the thing. The problem was the distance between seeing it and building it.

There is a name for that failure, and it turns out it is not a personal one.

A gap with a name

Twenty-five years ago, two Stanford professors, Jeffrey Pfeffer and Robert Sutton, wrote a book about it. Their subject was companies: organizations that knew exactly what to do, could describe it in detail, could even present it on a slide, and still could not bring themselves to do it. The knowledge was present. The action was missing. The space between the two was where strategy quietly went to die. They called it the knowing-doing gap.

What struck me, reading it long after it was written, is that the same gap operates at the scale of a single person. For an individual it does not look like a stalled initiative. It looks like a notebook full of starts, a dozen half-built things, and the recurring experience of watching the market catch up to a thought you had two years earlier. The mechanism is identical. Knowing what to do was never the hard part.

Knowing was never the scarce resource. Doing was.

For most of modern history we built our institutions around the opposite assumption. Information was expensive, so we treated knowing as the prize and execution as a detail that would sort itself out. Degrees, credentials, access to research: all of it optimized for getting people to know the right thing. Almost none of it addressed the part that was actually rare, which was the capacity to convert knowing into a finished, shipped, in-the-world result.

The real bottleneck was activation energy

For idea-rich people, the constraint was never imagination. It was activation energy, the cost of starting. The blank page. The working memory that drops a brilliant thought before you can get it written down. The unglamorous hour of setup that sits between seeing a thing and having the first rough version of it in front of you, where there is nothing to react to, nothing to fix, only the friction of making something out of nothing.

That hour is where my best ideas went to die. Not because the ideas were weak, but because the distance between "I see it" and "I built it" was just long enough for the spark to cool. By the time the page was no longer blank, the energy that made the idea feel urgent had drained off, and the thought rejoined the pile of things I would get to eventually. Multiply that by a career and you get a very specific kind of quiet loss: not the ideas you never had, but the ones you had and never moved on.

What actually changed

Here is the part I think matters more than the productivity numbers everyone is arguing about. AI is not valuable to me because it has ideas. I have never been short on ideas. It is valuable because it collapses the distance between a spark and a tangible thing before the spark dies.

It catches the thought. It kills the blank page. It hands me a rough draft in ninety seconds, and reacting to a draft is easy. Editing is comfortable. Judgment is comfortable. Telling something that is almost right why it is not quite right is the work I was always able to do. Starting from nothing was the wall, and the wall is what stopped me.

The wall is gone.

Notice what this is and is not. It is not that the machine supplies the insight. The insight, the taste, the sense of what is worth making and what is not, still has to come from a person. What the machine supplies is the first ugly version, the scaffolding, the momentum. It moves the starting line from "nothing" to "something to fix," and for a certain kind of person that single change is the difference between a finished thing and one more entry in the notebook.

The loop

The piece you are reading almost did not exist. The idea behind it landed the way thousands have before, in a spare moment, and under the old rules it would have vanished within the hour. The difference now is a simple loop: capture the spark the instant it hits, let AI turn it into a rough draft, then react and ship. Catch it, shape it, finish it. That is the whole trick, and it is not complicated.

None of those steps is new on its own. People have always captured ideas, drafted, and revised. What is new is that the most expensive step, getting from a blank page to a workable first draft, has collapsed from hours to seconds, and it is precisely that step that used to break the chain. Remove the break and the rest of the loop, which most of us could always do, finally runs end to end.

The honest caveat

I sell clarity, not hype, so here is the uncomfortable other half. AI can also become a very satisfying way to feel productive while shipping nothing. Generating, refining, exploring, regenerating one more time: it all feels like work, and it produces the warm sensation of progress without the result. The tool that removes the wall can just as easily build a very comfortable room on the near side of it, where you spend the afternoon polishing something you never intend to release.

The discipline of finishing still has to be yours. The machine will draft forever if you let it. It will not decide that a thing is done, defend a point of view, or carry the small risk of putting your name on something and letting other people see it. Removing the wall does not remove the need to actually ship. It just means that when you fail to ship, you can no longer blame the blank page. The excuse is gone along with the wall, and that is uncomfortable in a useful way.

Why this is bigger than a feature demo

Step back from my own desk and the implication is larger. An entire class of people whose ideas used to vanish in the gap can now act on them. For years, the ability to execute quietly sorted who got to create and who only got to imagine. It rewarded a specific temperament, the people who could push through the cold-start friction, and it left a great deal of good thinking stranded in the heads of people who had the ideas but not the activation energy.

That constraint is loosening. The result is not that everyone becomes a genius. It is that the bottleneck shifts from starting to judgment, from "can you get it built" to "is it worth building and will you stand behind it." Those are better problems to have, and they reward a wider range of people than the old ones did. That is the real story of this moment, and it is bigger than any single product launch or benchmark score.

If you have ever watched someone else build the thing you thought of first, this one is for you. The wall that stopped you was real, and for the first time in a long time it is not there anymore.

So the question I would leave you with is the one I had to ask myself. If the wall is gone, what is the idea you have been carrying, the one you keep watching other people build, that you could finally act on this week?

At Stewart Consulting, that is the work I care about: helping teams turn AI from a talking point into shipped work, with the trade-offs left in and the discipline kept intact.

An independent perspective from Stewart Consulting, published for discussion. Questions or feedback are welcome. See Connect. The concept of the knowing-doing gap comes from The Knowing-Doing Gap: How Smart Companies Turn Knowledge into Action by Jeffrey Pfeffer and Robert I. Sutton, Harvard Business School Press, 2000.