When an Idea Meets Claude Max

When an idea meets Claude Max, it runs like The Flash.

I wrote that line on the fourth day. Not as praise, but as a note from someone still taken aback. I have spent more than a decade writing code around a day job, so I know the rhythm by heart: the idea arrives today, the prototype takes shape next month—if I find the time at all. That rhythm collapsed in a matter of days.

One Person, and It Feels Like Having Several Employees

What struck me was not the typing speed, but the disappearance of the obstacles to execution. An idea used to stall because it had to wait behind research, behind boilerplate, behind documentation I had not gotten around to reading, behind a bug that took two late nights to trace. Every one of those delays carries a cost—not a monetary cost, but a cost in morale. Most of my ideas died not because they were wrong, but because they ran out of breath along the way.

Now all of the work that used to block the path can be done in parallel. One person feels like having several employees who never tire, never need a meeting, and never take offence when I throw their work away and ask them to start again from zero. An idea that realistically used to take months can, given a well-made plan, be finished in days—and the result often exceeds my own expectations.

The key phrase there is “given a well-made plan”. This is the part people rarely discuss. However fast the tool, all it does is magnify what is already in our heads. If all that is in there is fog, then fog is what comes out—only a version executed very neatly and very quickly.

Why I Am Beginning to Understand Those Who Walked Away

Those few days were enough to give me a better understanding of why some AI researchers have chosen to step away from their own work.

My guess—and it is purely a guess, not a claim about anyone else’s mind—is that the question is no longer simply about pushing the limit. If it were only about a ceiling, that ceiling would still feel like ours: we decide how far to go, and we are the ones applying the brakes. The reality feels different. It is as though something has crossed the line of being “merely a creation”, and the maker has ended up a spectator to the work of their own hands.

I am not saying this machine is conscious or alive. I only want to say this: the sense of being in control turns out to be far more fragile than I imagined. Four days were enough to shift that feeling in me, and I am nothing more than a user. I cannot imagine what it is like to be someone who sees the next version of the AI every day, before the world gets a glimpse of it.

What Is Scarce Now Is Not Execution

The entire digital economy has stood on a single assumption: execution is expensive. “Ideas are free, execution is what costs”—this is the old advice we keep repeating. Everything built on top of it follows that assumption: startup valuations, what developers charge, the advantage held by large companies, even the respect we extend to those who “managed to build something”.

That assumption is now being pulled away. If the price of execution falls close to zero, only three scarce things remain: taste, distribution, and a reason. Why does this product need to exist? For whom? And why now?

A Saturation Point That Appears Unavoidable

For that reason, I am fairly confident a saturation point is coming, and before long. Content, SaaS, applications, whatever it may be—all of it will flood the market until human attention can no longer hold any of it.

This is not an apocalypse that burns, but one that drowns. Water rising slowly until everything looks the same height, and nothing stands out any more.

Honestly, though, I have to leave room for the possibility that I am wrong. Every wave of technology has brought a similar prophecy of saturation: the printing press, blogs, the App Store, YouTube. What usually happens is not total saturation but a shift in standards. The floor of quality rises; whatever is merely “average” becomes worthless, and the real value runs to the two ends of the spectrum: the genuinely excellent, or the genuinely personal. Perhaps it will be the same this time. Then again, perhaps not, because what is flooding the market this time is not people, but machines that never sleep.

Cipta, Rasa, and Karsa

The part that troubles me is not the flood of products, but what is happening to the people.

Dependence on this tool arrives far more quietly than I expected. Not in the form of “I cannot work without this tool”, but in the form of a pause that has gone missing. There used to be a pause between “not knowing” and “coming to know”—an uncomfortable pause, the place where we force the brain to work hard. That pause can now be skipped at any time. And the brain, like water, always takes the path that slopes downward and comes easiest.

This is the condition I would call: satisfying rasa without karsa. Javanese has three words for a whole human being: cipta, rasa, karsa. The power to think, the power to feel, and the power to will.

• Cipta is slowly being handed over to the machine. We stop composing, and merely choose from the options put in front of us.
• Rasa is indulged instead. The result is tidy, fast, pleasant to look at, and satisfying at once.
• Karsa (the will) is the one that bears the loss most quietly. Will grows out of friction and difficulty. If there is no friction left, will has no room to form.

In the end, what remains is a person who is satisfied but wants nothing. A critical stance becomes expensive, because questioning things is tiring, while accepting what is handed to us is instant and already good enough. That is where creative power erodes—not because it is forbidden, but because it is no longer needed.

So, Will I Stop Using It? No.

Refusing this tool would be as naive as refusing electricity for fear of growing soft. I will keep using it every day. What I have changed is not the tool, but my own house rules:

1. Write the plan by hand first. Before sending a single prompt, I force myself to write down what I want to build and why. If that part is delegated as well, everything else follows it out the door.
2. Deliberately preserve the pause. Not everything that can be sped up deserves to be. For the things I genuinely want to master, I have to try until I am stuck first, and only then ask the AI.
3. Watch the learning ratio. If I produced a great deal this week but understood nothing new, that is a warning signal—not an achievement.
4. Chase the reason, not just the output. Output is cheap now. The only thing still setting anything apart is the reason it needs to exist at all.

Ideas really do run as fast as The Flash now. But speed is only useful if we still know where we are headed—and still have the will to choose that direction ourselves.