You’re about to graduate into a market with jobs that don’t quite exist yet, holding a $200k piece of paper that means less than it should. And at the same time, you’re expected to walk into a company and be the most AI-fluent person in the room. The same institutions that spent your last few years treating AI like a form of cheating are now handing you off to employers who assume you've got it figured out.
I teach undergrads, and recently a student told me he’d gotten in trouble for using AI to outline a report that he then wrote entirely by himself. That’s the bind you're in. You're expected to be fluent in something you were told not to touch.
Even if you wanted to learn every AI tool out there, you would never stop chasing. The tools change every day. Becoming an expert at a particular vibe coding tool or learning to write the perfect prompt for this month's model can mean pouring your time into a skill that’s obsolete by the time you actually start the job. The half-life on a lot of this stuff is measured in months now. So which tool you learn matters a lot less than the one skill underneath all of them: the ability to teach yourself whatever comes next.
What Is the Most Important Skill for New Grads in Today’s Economy?
Agency — the proactive, self-teaching drive to solve unfamiliar problems — matters more than specific AI technical skills because the tools change constantly, rendering specific training obsolete within months. Employers prioritize candidates who demonstrate the instinct to figure things out independently over those with prestigious degrees or temporary technical expertise.
Build Agency, Not Specific AI Skills
If I could showcase one thing on my resume, it would be agency. In the usual sense, agency is about self-determination, your capacity to make things happen instead of waiting for them to happen to you. Aimed at whatever problem is in front of you, it’s the willingness to pick up something unfamiliar and run with it. To try. To take the first step toward solving something you’ve never seen before without waiting for someone to teach you how.
That’s the skill that doesn't expire. A specific tool will be obsolete in a year, but the disposition that lets you learn the next tool and the one after that lasts your whole career. If you treat specific AI skills as fixed things to acquire, you’re aiming at a target that keeps moving. Agency is what lets you keep up with the economy, however fast it moves.
What Agency Looks Like
I’ve mentored many students with varying levels of agency, and it shows up in their work. One student would hit a roadblock while prototyping and then wait a full week to tell me it hadn’t worked and ask what to try next. At some point, I started wondering if I was spending more time on their project than they were.
Another student, given the same minimal instructions, came back with a fully designed app with a differentiated concept, edge cases handled, a brand kit, and a financial plan for investors. The assignment hadn’t asked for most of that.
The difference between those two students — waiting for someone to unblock the project for you versus going well above and beyond the requirements you were handed — is agency. And it’s the quality I see that most consistently separates the people who are thriving from the ones who feel stuck.
People doing the hiring are starting to say this exact thing. Entrepreneur Phoebe Gates recently argued that the gap between a brand-new grad and a senior hire is shrinking. The real differentiator is whether or not a person has high agency. Her company asks candidates how they use AI in their own lives, not because the job really requires it, but because it’s a signal.
She also pointed out that you don’t really need to be technical anymore to use AI well. The person who set up their own AI agent at home with nobody asking them to is showing exactly the proactive, self-teaching instinct they want. The advantage now belongs to whoever has the instinct to go try, degree or not.
I heard nearly the same thing in a hallway conversation with a woman who runs a company that generates films with AI. She’d rather hire someone who shows up with their own ideas and the drive to figure out how to build them than someone with the technical chops who waits to be handed a problem. The skills, she figures, they can pick up. The instinct to chase down their own idea is the part she can’t teach. That instinct is agency.
And Max Schoening, the head of product at Notion, recently made the case that cultivating agency matters more than cultivating skills in the AI era. What carries you now is the willingness to figure things out when nothing is stable.
The Opening Nobody’s Talking About
There’s something I find freeing in all of this. Agency isn’t gated behind a degree, a GPA or four years and a pile of debt. It’s available to you right now, for free.
You’re not stuck behind a group of people who have it all worked out. Plenty of experienced professionals feel exactly the way you do, unsure what the next few years hold and carrying their own kind of debt, just in the shape of mortgages and medical bills instead of student loans.
Nobody is fully prepared for a workforce that’s still being built in real time, and no degree, however impressive, gets you out of having to navigate it for yourself.
If you want somewhere to start, it doesn't have to be ambitious. Pick the most mind-numbing task in front of you, the one you quietly dread, and teach yourself to make it disappear. I’d bet you'd learn more from automating one tedious spreadsheet than from any course promising to make you AI fluent.
The first tool I built for myself wasn’t even about productivity. I made an agent whose whole job was to make me feel more human at work: remind me to drink water, log off, reflect on something good and turn my rambling, out-loud thoughts into organized plans. Next, I automated the thing I’ve dreaded my whole career: writing research reports with standardized formatting. That one didn’t feel like efficiency so much as relief. I no longer had to sort my thoughts into an acceptable shape before I was allowed to share them.
When something you build works, share it because the odds are good that someone else’s brain works like yours and needs the same thing. Drop it in the Slack, Reddit or Discord channels where people are already hunting for solutions. Tell your classmates. Write a short post on how you made it or record a two-minute screen capture so someone can build on top of it. The format matters less than the fact that you put it somewhere public instead of letting it die in your downloads folder. And once other people start improving it, you get to use the better version too.
Take the same instinct to the job search itself. I’ve personally never once gotten an interview without a referral. Message strangers on the internet to ask them about their jobs. Most of them will ignore you, but the one who doesn’t can change everything.
Build the Future You Actually Want
Instead of another “AI is the future” speech, I’ll leave you with one question. Start with the world you actually want to live in. What does it need, and what would you build to bring it closer?
My answer is that I’d build things that make us feel more human. Yours will be different, and you don’t need it figured out by graduation. You just need to be the kind of person who goes looking for it.
