Your AI credentials don't matter. Your artifacts do. Here's how to build both sides — plus I'm launching Nate's Network.

Nobody is admitting this publicly, so I will: there are essentially infinite AI jobs right now. Not “growing demand.” Not “a hot sector.” Infinite, as in every company I work with has an uncapped budget for AI talent and will hire as many qualified people as they can find, and they still can’t find enough. I’ve been kicking around tech for decades and I have never seen a labor market this lopsided. Accenture is training 700,000 people on agentic AI. The ManpowerGroup survey of 39,000 employers across 41 countries says AI skills are now the single hardest capability to find on Earth. Not among the hardest. The hardest.

And nobody is breaking down the specific, learnable skills you actually need to get those jobs.

I dug through hundreds of real job postings at companies hiring right now: Anthropic, Robinhood, Upwork, Glean, Scale AI, and dozens more. I found seven skills that show up consistently across all of them, whether the title is AI Engineer, AI Product Manager, Agent Operations Lead, or domain specialist. They are specific. They are learnable. And most people already have two or three of them from whatever they’re doing right now.

If you’re reading this thinking “that’s not for me,” it’s extra for you. We are short people for the AI economy. You don’t need a CS degree. Paul Graham and Sam Altman keep saying the magic word is “taste.” Poggio cofounder Matt Slotnick called that out: “The taste thing works because it’s nebulous, unassailable, and it feeds the ego.” He’s right. Taste is not a skill. But the seven concrete things underneath it are.

The reason the gap keeps widening despite the ocean of “AI upskilling” content is that almost none of it maps to what employers are actually hiring for. It’s taught at the wrong altitude: “AI for Everyone” at the top, deep ML engineering at the bottom, and the entire middle layer barely exists as curriculum. It teaches tools when employers are hiring for judgment. Most courses show you how to use ChatGPT; employers are hiring for “agentic evaluation mindset, including automated evals, simulation-based testing, regression frameworks, and metrics design.” And the credentials are mostly worthless. DataCamp surveyed 500 enterprise leaders this year: 40% of AI training is video courses, 23% of leaders say it doesn’t translate to real work. As recently as 2024, Accenture’s own data showed only 26% of workers had received training on how to actually collaborate with AI. The certification landscape is mostly badges from Saturday afternoon lectures that hiring managers have been ignoring for years.

So I’m building something different.

Here’s what’s inside:

• The K-shaped split. Why “infinite jobs” and “I can’t get hired” are both true at the same time, and which side of the split you’re on.

• The hiring side is broken too. A meaningful percentage of AI job postings aren’t real job postings, and a code of conduct for fixing it.

• The seven skills that pay $150K–$400K. Extracted from real postings at Anthropic, Robinhood, Upwork, Glean, and Scale AI, mapped to four career tracks with a twelve-week learning path using entirely free materials.

• Nate’s Network. I’m launching a vetted AI talent network that is built to match the right people to the right roles faster, based on what they have actually built and trained on.

The labor market has a structure that most people can’t see. Once you can, the path through it gets specific.