The résumé stopped being proof the moment both sides started using AI.

TL;DR: Résumé screening worked for decades because tailoring a résumé to one employer cost the candidate real effort, and that effort, not the wording, was the signal you were reading. AI dropped the cost to zero, so a perfect-looking match now tells you nothing about who sent it. You can't get that signal back by reading harder. You buy a new one, by making the application itself cost a candidate something only a serious one will pay.

Most founders still relate to hiring the way they related to it five years ago. Post the role, résumés come in, you read them, the strong ones rise, you interview the top handful. Somewhere in there is a filter doing real work, separating people who fit from people who don't.

That filter is broken, and it broke quietly enough that a lot of operators haven't noticed yet. My partner Ryan Deiss described what it looks like from the inside at Scalable: you post a role, and the résumés that come back look suspiciously perfect. Every past job lines up with what you asked for. Every phrase sounds like it was written specifically for you, because it was. Then you get on the phone and the fit isn't there. The résumé said everything right. The person is a stranger to the role.

This isn't a story about one bad hiring cycle. It's what happens when a signal that used to require effort to fake now requires none. Writing a résumé that speaks the employer's exact language used to take real skill, or real desperation, and either one told you something. Now it takes one prompt. The people who would have washed out at the résumé stage, because they couldn't be bothered to tailor their pitch, are indistinguishable on paper from the people who actually fit.

The instinct is to read harder. Tighter keywords. A sharper applicant tracking system. More questions on the form. All of that assumes the signal is still somewhere inside the document, buried under polish, and that enough scrutiny will surface it.

It isn't in there. What you were reading was never really the content of the résumé, it was the price of producing it. The afternoon a candidate spent rewriting their history in your language was the part that told you they wanted the job. The words were just the receipt. Drop that price to zero and there's no receipt left to read, however carefully you read it. Every résumé in your pile now costs the same to send, which is nothing, and a stack where everything costs the same carries no ranking information at all. Forbes, covering a Robert Half survey of more than 2,000 U.S. hiring managers, reported that 67 percent of them say reviewing AI-generated applications has slowed their hiring. That gets read as a volume complaint. It's narrower than that. Hiring slowed because the step that used to sort the pile for free stopped sorting anything, and the sorting moved onto people's calendars.

If what you lost was a cost the candidate used to pay, the only way to get it back is to charge again, somewhere other than the résumé.

Ryan's team at Scalable runs four versions of this, and none of them are about reading better. The application asks a question so specific to a real situation that AI polish struggles to answer it convincingly, because there's nothing generic to draw from. Serious candidates, even at the executive level, get asked to do a small piece of paid work, whether that's a CFO building a real 90-day plan or a marketing hire working through a problem live on a ninety-minute call, because watching someone think for ninety minutes tells you more than reading their history for five.

The job posting itself ends with an honest paragraph about what won't be fun about the role, on the theory that a wrong-fit candidate reads it and quietly leaves, sparing everyone the interview. And before anyone gets to a live conversation, Scalable sends one more message, from an assistant, restating the pay and every downside already listed, with an explicit invitation to opt out. Roughly a third of people do.

Not one of those four is a judgment you make. Each one asks the candidate to spend something: an hour of real thinking, an honest look at what they'd dislike about the job, a decision to stay in after being told the worst of it. Three of the four land before anyone has invested much of anything, and the paid work sample comes later, for the people who cleared them. Someone who wants the role pays without hesitating. Someone who fired off forty applications before lunch doesn't, because paying was never part of the plan. The pile shrinks before anybody judges anything, and what's left in it has a price tag attached again.

The founders still relying on the old filter aren't lazy. They're paying a tax they can't see. Every hour spent parsing a stack of AI-perfect résumés, every interview burned on someone who applied without reading the role, every hire that looks right on paper and unravels in ninety days, is the cost of trusting a signal that stopped carrying information. It shows up as slower hiring, worse hires, and a founder who's now personally back in the interview chair because nobody else can be trusted to sort the pile. That's the opposite of leverage. It's the founder doing, by hand, the one job that was supposed to have been automated away by a process.

It also compounds. A bad filter doesn't just cost you a bad hire now and then. Once a résumé can't be believed, every round after it has to do the work the résumé used to do, and every one of those rounds costs something. Notice who's paying. You are. The point of a filter was that the candidate paid to get through it. Leave it broken and you quietly pick up the bill.

The résumé wasn't a filter you built. It was a filter you inherited, and it worked for decades because both sides played it straight. That era is over, and it isn't coming back regardless of how good your applicant tracking software gets.

What replaces it is older than the résumé and a great deal harder to fake: hire on what a person was willing to spend, not on what they were willing to claim. Claims are free now, permanently, and no software is going to make them expensive again. Spend never gets cheap. So stop asking candidates to describe themselves well and start asking them to pay a little for the chance, then hire the ones who did. The founders getting this right this year aren't reading résumés more carefully. They stopped treating the résumé as the place where anything gets decided.

— Roland

The A-player with the big-company résumé is the most expensive hire you'll ever make.

I've heard this hiring advice more times than I can count. Go get the one with the brand name on their résumé and the track record at scale.

It sounds right. It's a trap.

That person was shaped by an environment with an unlimited budget. Ask them a simple question and their instinct is to staff up, because that's how their old company solved everything.

You're not buying their judgment. You're buying the habits of a company that never had to choose.

Want the full breakdown? Read the post.