Over the last four years, the Pear Talent team has made more than 300 hires for pre-seed and seed founders. For most of that time, we had a pretty repeatable playbook for who to hire at this stage and why.
The explosion of AI coding tools broke that playbook. You’ve probably read a dozen versions of the same hot take since: “We’ll never hire junior engineers again.”
That take was wrong, and new grad hiring has quietly crept back into the startup ecosystem. But not for the reason you’d expect.
Founders used to pay a premium for 2 to 4 years of experience, because that experience was cheaper than a senior hire and could still ship independently. AI made that premium pointless, and everyone assumed the answer was to just hire cheaper. New grads looked like the obvious swap. But that doesn’t mean the talent got cheap. It means the market is currently mispricing new grads as a discount option when they’re actually the best deal on the table, and that mispricing won’t last.
So what are founders actually paying for now?
Why experience used to be worth paying for
There used to be a real gap between a new grad and someone with two to four years of experience.
You hired the experienced person because they shipped faster, wrote cleaner code, and needed less hand-holding on stuff that already had a known answer. You could hand them a scoped problem and walk away. That was the deal.
You were paying for speed and finished problems, not potential.
As coding tools closed that gap, the real question became whether founders needed to hire the same way they used to. The answer was yes, but not in the way most people ran with it.
Here’s the trap founders are walking into right now.
Series A teams that used to run 12 to 16 people are now closer to 6. The instinct to run lean is correct, and it’s already reshaping who founders look for. But the assumption underneath it is the trap: that because the execution gap closed, the talent itself got cheap.
It didn’t. The question was never how many people to hire. It’s which kind of person you’re still willing to pay for
What you were actually paying for: Judgment vs. execution
Break the experience premium into two things:
- Judgment: Knowing what not to build. Picking an architecture you won’t regret in 18 months. Debugging the thing that’s on fire in production at 2am when the logs make no sense. That’s still scarce, still expensive, and AI hasn’t touched it.
- Execution: Scoped, known-pattern work, delivered fast and clean. That’s most of what a 2 to 4 year hire was actually getting you. And that’s the exact slice Claude Code and Codex now do for almost nothing.
So if you’re still paying the 2 to 4 years of experience premium for execution and speed specifically, it likely is not worth it. The thing that’s actually still valuable, judgment, was never what that premium was buying you in the first place.
New grads: Hire them before someone else does
While all of this was happening, there was a deep pool of strong new grads sitting available. Three years ago, most of them would have taken a name-brand offer without a second thought. Then Big Tech pulled back hard on entry-level hiring, and that pool just sat there.
With a market full of high-slope candidates and nowhere near enough of the old competition for them, the premium founders used to pay for experience stopped making sense.
After hiring a good number of these candidates and watching what happened next, the real advantage wasn’t that they matched their more experienced peers on day one. It’s what compounded after. They grew up inside your codebase. They knew your customers firsthand. They absorbed your product instincts because they built them there, not somewhere else first.
That slope happens at your company, on your dime, early in your build-out. Wait two years for that same person to prove themselves elsewhere, and you’re not making a hire anymore. You’re stepping into a bidding war for someone who’s already proven, already expensive, and already fielding five other offers.
Hiring new grads only works if you actually develop them
The compounding isn’t automatic. You have to make it happen.
A new grad only turns into a high-slope, deeply embedded asset if someone actually invests in them. They need real mentorship, real ownership, not ticket-shuffling, and work that’s scoped enough to finish but stretchy enough to grow on. A lean team is exactly the place where that kind of slack is hardest to find, because everyone’s already underwater.
So be honest with yourself before you make the hire. Do you have someone who will actually spend time developing this person? Is there work that builds them, not just work that needs doing? Can you tolerate a ramp that’s real and measured in months, not days?
If the answer is no on any of those, don’t make the hire. You’ll get a bad outcome, the person will struggle, and you’ll walk away having confirmed the exact bias this piece is arguing against.
How to actually screen, onboard, and pay for this
Here’s what actually matters when hiring new grads.
Screen for slope, not pedigree
You’re not buying what they know today, you’re buying how fast they climb. Don’t ask “walk me through your resume.” Ask questions that force them to show you a moment of real difficulty:
- “Tell me about something you built or learned that was genuinely too hard for you at the time. What did you do when you got stuck?”
- “What’s something you taught yourself that nobody assigned you?”
- Give them a small, unfamiliar problem live and just watch how they work through not knowing the answer. You’re not grading the output. You’re watching how they close the gap.
Pedigree tells you where they’ve been. Slope tells you where they’re going. Hire the slope.
Win on access, not brand
You can’t out-prestige the offers your competitors used to make. So don’t try. The best early-career candidates right now aren’t optimizing for a logo, they’re optimizing for how fast they’ll grow. Sell them on what you can actually offer: direct access to the founders, ownership of something real in month one, and a growth curve no big company can match.
Give them a real 90-day plan, not a vague “we’ll figure it out.”
Something like this:
- Weeks 1-2: Small, well-scoped task with a visible finish line. Something that ships. The goal isn’t impact yet, it’s building the muscle of shipping something end to end at your company.
- Weeks 3-6: One real project with actual ownership, paired with a specific person who’s responsible for unblocking them and reviewing their work. Not “ask anyone,” a named person.
- Weeks 7-12: Something with real stakes, tied to a metric the company cares about. By the end of month three, you should know whether the slope is real.
If you can’t sketch this out before you hire, you’re not ready to make the hire yet.
Let equity do some of the work
You won’t always win on cash, and you don’t need to. What you’re actually offering is ownership at the point of maximum upside, to someone with 30 years of career ahead of them. Say that plainly instead of burying it in a comp band. It’s the same pitch you make to your own investors, just handed to the person who’s going to help build the thing.
On a 6-person engineering team, you’re not hiring a class of twenty new grads. You’re making one or two bets. Get those right and they move the whole company. Treat the decision with that much weight.
Buy the slope
Experience got cheap because the tools made it cheap. Trajectory stayed expensive because no tool can manufacture a person who compounds inside your company for years. The founders who see that gap now and act on it will lock in the talent everyone else will be fighting to poach two years from now.
Buy the slope while it’s still on sale.