This morning, before my first coffee, I made the mistake of opening my phone.

A business mentor, the kind with a ring light and a lanyard at every conference, was telling a founder that his accounting SaaS would be obsolete within months. Then he kept going. Custom software development? Obsolete too. The only real products left, he argued, were LLMs (Large Language Models) and SLMs (Small Language Models). Everything else was a relic waiting to be automated away.

I put the phone down. I grabbed my coffee. I sat with it for a moment.

There’s a version of this argument worth taking seriously. AI has genuinely changed how software gets built, how decisions get made, how products evolve. We use it. We integrate it into the projects that need it, we combine it with custom development when the problem calls for it. So the trend doesn’t seem wrong to me. What seems incomplete is what it leaves out.

These tools don’t operate themselves.

Behind every model, every workflow, every AI-integrated platform, there is someone who decided what problem it should solve. Someone who understood the client well enough to know what “better” actually means in their context. Someone who made the call when the output was technically correct but contextually wrong. That person, with their experience, their judgment, their specific way of seeing a problem, is not a feature you can prompt your way into.

The Irreplaceable Value of Human Judgment

There’s a word in Spanish that doesn’t translate cleanly: “maña”. It means the particular skill and know-how a person develops through years of doing something specific, the kind of mastery that lives in the hands and the memory rather than in any manual. A carpenter’s exact chisel angle. A senior engineer’s instinct about where a system will break before it does. A grandmother’s decision about when to turn off the heat.

AI as the New Industrial Revolution

The industrial revolution changed how things were made. It didn’t change what made things worth having.

AI is this generation’s industrial revolution. It compresses effort, accelerates output, democratizes access to capability that used to require years of specialization. That’s real, and it matters. But it also raises the ceiling on what the humans behind it can achieve. It doesn’t replace them.

Every client we work with has a context that no model has seen. The value is in knowing how to ask the right questions before writing a single line of code, and knowing what to do when the answer surprises us.

This is not an argument against AI. It’s an argument for understanding what it is: a powerful tool, shaped entirely by the judgment of the people who wield it.

Tools change. What doesn’t change is that someone has to know when and how to use them. If you want to see how we do it, with AI, with custom development, or with both, it’s all at inflexionsoftware.com.