Advantages in the Age of AI

I used to think technical skill was the cleanest moat. It was certainly the most straightforward: learn more, build more, intern more.

But AI is making that belief harder to hold. “Intelligence” is getting cheaper and more abundant. AI may not yet replicate the brightest minds, but it can already handle parts of many junior roles: coding simple websites, doing market research, answering sales calls.

And that changes the advantages. If more people can build the first version of a product, the more important question becomes: are they building the right thing, for people who actually care?

Technical implementation alone is becoming less defensible. But deep technical judgment is far from obsolete. We still need to architect well, have good product taste, and know when AI is wrong. We no longer need to go as deep to know how to build things from scratch - we just need to know enough to 1) ask the right questions and give the right prompts so AI can do it for us, and 2) verify that AI isn’t giving us slop or BS.

So what are the moats in 2026?

Moats imply asymmetry: having something others do not. I increasingly think the moats are - and I increasingly believe the moats lie in speed, access, and trust.

Speed matters because it shortens the time from guesswork to correction. The faster we ship, the faster reality tells us when and where we’re wrong. We see dogfooding and shipping as research. Importantly, onboarding customers earlier gives us better feedback and data - creating a learning loop where products are iteratively improved and models are constantly retrained.

Access matters because real problems aren’t always visible from the outside. Anyone can ask AI for startup ideas; fewer people are standing close enough to real pain points to know what problems matter.

A doctor building software for clinical workflows has an advantage that a technically stronger outsider may not have. The doctor knows which part of the workflow is actually annoying because it personally impacts him.

I’ve always believed success is a product of luck and skill. Having access makes luck less random. Being in the right room, knowing the right people, or having worked inside the right industry at the right time exposes one to problems before they become obvious to everyone else.

Trust may be the hardest moat for AI to touch. AI can write a sales email, suggest a GTM strategy, and generate a list of leads. It cannot inherit our reputation. It cannot make warm introductions. Trust building is still a human process - takes a human to trust another human.

That’s why two companies can have access to the same AI tools and still get very different outcomes. The tool can help both of them execute. But it can’t give both of them the same credibility or customers.

I still think technical skill matters. But it’s no longer enough by itself. The deeper advantage is knowing what to build, knowing how to sell, and knowing the right people (whether its your team, customers, VCs, etc.). And that’s as uncomfortable as it is challenging, because while technical skill is something that feels controllable, access and trust are messier.