🎧 Listen to this article (4:53)
Knowledge is becoming cheap. AI can research, explain, compare, plan and make recommendations that once required specialist expertise.
Execution is getting cheaper too. Agents can already write software, analyse data, produce documents and carry out complex work.
That changes where value sits.
Knowing and doing are no longer enough
In Knowing Which Screw to Turn, I wrote about judgment: understanding a problem well enough to know where intervention matters.
That still matters, but AI is moving into that territory too. A model can suggest which screw to turn, explain why and offer alternatives. An agent may soon turn the screw, too.
So the valuable part cannot simply be knowing or doing. Someone still has to care what happens afterwards.
Reasoning and execution are capabilities. Responsibility is a relationship.
It means staying with the problem when reality disagrees with the plan, deciding what happens next and finding out whether the work actually helped.
Outputs are easier than outcomes
A consultant produces a strategy. A designer produces a website. A developer produces software.
Those are outputs, and professionals organise work around them for good reason. Outputs are relatively easy to define, control, deliver and invoice.
Outcomes are messier. A designer cannot guarantee that a website will produce sales because too many variables sit outside one person’s control. But that does not mean the outcome stops mattering.
The full-stack mentality keeps asking the question after delivery: what was this supposed to change, and did it?
Responsibility does not mean promising results you cannot control. It means owning what you can control, staying engaged with what you can influence, and learning from what you cannot.
The last mile is part of the work
A good idea can still fail because nobody implements it properly. The client gets stuck, circumstances change, an assumption proves wrong, or the final ten per cent has more friction than expected.
That was the point of The Last Mile. Almost finished is not finished.
The full stack goes one step further. Finishing the output doesn’t finish the problem.
You have to see what happens next. If it worked, understand why. If it did not, adapt. If something broke, fix what you reasonably can and check again.
That is where responsibility turns execution into an outcome.
Prevention is part of the stack
The fire and rescue service makes this unusually easy to see.
Its success cannot depend on maximising the number of fires it attends. Education, inspection, auditing and prevention are valuable precisely because they reduce the need for rescue.
An audit that identifies a dangerous problem and produces a report is still only an output. Prevention happens when somebody acts on the warning, fixes the problem and verifies that the risk has actually been reduced.
A warning that nobody acts on is not prevention.
The interesting part is the incentive. The fire service can pursue fewer fires because its success is not measured by how many disasters it can bill for.
That raises an uncomfortable question for any business: what happens when your revenue depends on the problem continuing to exist?
Do not manufacture dependency
Once a problem has been solved, how much capability should remain with the client?
There is nothing wrong with someone choosing a managed service. I do not need my plumber to teach me enough plumbing that I never call one again. Sometimes I simply want somebody competent to handle the problem.
Chosen dependence is not the issue. Manufactured dependence is different.
If a system is deliberately kept mysterious, undocumented or unnecessarily difficult so the customer has no choice but to return, recurring revenue comes from preserving weakness.
There is a better model: solve the problem properly, explain what matters, document what should be understood and leave behind a stronger system and a more capable client.
That is recurring business through progression, not recurring business through dependence.
Good systems reduce the need for rescue. Good education reduces unnecessary dependence. That does not make expertise worthless; it changes what expertise is for.
The goal is not to remain necessary. The goal is to remain valuable.
The whole stack
I have been circling different parts of this idea across several pieces.
Knowing Which Screw to Turn was about judgment. The Last Mile was about execution. Accountability Is What Happens Next was about responsibility.
The full stack connects them and adds the final step: verify that something actually improved, then transfer enough understanding so the improvement can survive without you.
Run the race with them. Carry them if you have to.
Then help them become a stronger runner.
