← Blog

Selling the Recipe

Henry Kim ·

Selling the Recipe

At the end of coding automation, where do developers go? When an era arrives where apps are built with a single line of code, will all the efforts and experiences accumulated so far vanish like a mirage? This has been a recurring question for me recently.

No matter how smart an AI agent is, it has its limits. "General input produces general output." The reason is simple: The real know-how in an industry isn't a formula for success, but the accumulation of failures. Anyone can say what features would be good to have. But 'doing it this way led to failure in this industry' is something only those who have personally shed blood know. AI learns success patterns publicly available on the internet. However, it cannot learn about the failures quietly buried within the industry. That is the unreplicable domain. I am also experiencing this in the local field. While building apps for car dealerships, even identical features required hundreds of revisions to refine them to fit the industry's specific conventions. Without those calibration values, it would have just been an ordinary app.

Consider a chef. Using the same ingredients and tools, they produce entirely different flavors. Just like the saying, 'a secret recipe isn't even taught to one's daughter-in-law,' it contains not just a list of ingredients, but calibration values created through hundreds of failures. The IT market is now becoming the same. If someone with a recipe builds an app with AI, even using the same tools, a completely different result emerges. An era is coming where that 1-2% difference determines sales. History has shown this. When the internet gave everyone information, the value of information itself decreased, but the value of the ability to interpret information increased. If AI gives everyone coding ability, the value of coding itself will decrease, but the value of knowing what to build and how to build it will increase. It's not the person who writes the code, but the person who designs the recipe, who will survive.

Foloup · SKAIA research agenda · Talk to SKAIA