The Ceiling

The new old software economics

Generative Ai is a mind-twister. But it is also an economics twister. What used to take substantial time, effort and therefore money is now virtually free.

GenAi can create anything virtual if you just ask it to. Okay, it does not create long-form video, but anything from sites, processing pipelines and all the way to large software systems can be just prompted. If you can say it out loud, then you can build it. Now again okay… the cost is not really zero but it is going down fast.

Those who are systems thinkers know that every system has a limit to growth. Some systems also crash and burn when they grow out of bounds.

What will happen to software when the old limiting factor of coding gets removed? How will the new limiting factor emerge? What is the ceiling?

Knowing is the ceiling

I think the limiting factor is the knowing. It is knowing what is the right thing to do for the customer. It is knowing the customer deep enough. It is to anticipate the real needs of the customer better.

Knowing is building the right product!

We’ve always known more features is not necessarily better. Getting to more features faster is not the real game. System does not reward more features always.

Creating the right product is what the market, the system, rewards.

Doing it right is the other ceiling

Generative Ai is great at creating something matching the request. You say what you want and Ai delivers with output. But is it correct?

You can make the second prompt to the Ai: “Did you do it right?” Will you believe what the LLM says? How many times do you need to ask to be sure the LLM got it right. You see the problem.

Another sneaky ceiling starts to happen with quality. You can create anything very fast. The more you create, the more difficult it is to know it all hangs together. The more difficult it is to really know your system works right.

That is the second ceiling with generative Ai.

You see how this is so classic. Build the right product. Build it right. Validation and verification. Yin and yang.

Speeding up the loop. Giving oneself more odds at finding the needle

For the fun of it. Let’s dress up the rest into predictions

Strategic thinking will be more relevant than ever. Just because something can be imagined, should it be built? I believe the amount of startups will grow. The markets fragment into smaller niches. Those how can see the right “where to play” and “how to win” are those who thrive. Those who dare to make choices can still differentiate and succeed.

Architecture is another frontier that will be appreciated. We will re-learn how architecture either raises the ceiling or sets limits to growth. But the architecture will be something else than what we are accustomed to.

The real game has always been a continuous search and adjustment to the market realities. Those who have seen the ways to differentiate and to build the right capabilities have traditionally won. Market is a market still in the future. The game just speeds up.

Speed has always been of essence. Speed gives more learning loops. Speed gives the information on what works and what doesn’t the quickest.

It has always been about the insight to find the product. Righter is righter. The ultimate goal of GenAi in business rightness - not volume.

The frontiers that emerge

Antti -> Claude: You can write the last paragraph: Turn your viewpoints into next frontiers that the leaders in the industry will pursue. Match my writing style.

Claude: So what will the leaders pursue? Decision velocity becomes the new strategic muscle. Knowing what to build matters less than deciding fast enough to validate it in hours, not months. Quality transforms from a gate into a flow. Ship, observe, patch instantly - the LLM fixes as fast as it creates. The platform question looms large. Will we fragment into a thousand niches, or will one adaptive system serve them all? The answer shapes everything.

Claude: ⏺ Done! Added the closing section that turns those counter-points into forward-looking frontiers.