The Three Camps of Enterprise AI

Every conversation I have about enterprise AI eventually reveals which camp someone is in.

There are three.

The first camp is not worried at all. Their view: by the time we ship this, AI will have improved so much that today's problems won't exist. Give it three months. The model gets smarter, the guardrails get better. Just ship.

The second camp is also not worried, but for a different reason. Their view: technology has always disrupted and humans have always adapted. We went from assembly language to high-level languages and nobody stopped writing software. Developers didn't disappear, they just moved up the stack and did more interesting work. Let AI run its course. You'll find your place.

The third camp sits somewhere in the middle. They see the potential clearly. They're not dismissing it. But they also won't bet the house on any single assumption about where this goes. Novel technology comes with novel challenges and those challenges deserve real engineering attention, not a roadmap item that says "AI will fix this later."

I find myself in that third camp. Not because I'm pessimistic about AI. I'm not. But because I've shipped these systems and I know that "the model will get better" is not an architecture decision. It's a hope.

And here's the thing: none of them are entirely wrong.

Camp one is right that the technology is moving fast. Faster than most people's mental models of it. Things that felt unsolvable six months ago have real solutions today.

Camp two is right that humans are good at adapting. The assembly to high-level language transition didn't shrink the industry. It exploded it.

But camp three keeps me honest. We landed on the moon in 1969 with 4KB of onboard memory and software written by hand. No AI pair programmer. No LLM to generate the guidance code. Just engineers who understood every single line they were deploying and exactly what would happen if it failed.

In 2026, with all of modern technology available, we just flew around the moon again. The actual landing is still years away. Not because people aren't smart or working hard. Because getting hard things reliably right is still hard, regardless of how much technology you have.

More capability doesn't automatically mean more reliability. And in enterprise systems where a wrong payroll update affects someone's rent and a bad performance eval follows a career, reliability is the whole game.

The pressure to move fast is real. The optimism is warranted. The risks are also real. None of those things cancel each other out.

Which camp are you in?