The résumé is now noise

For most of my career, a résumé was a reasonable proxy. It was tedious to write, slightly painful to tailor, and that friction did useful work. A candidate who had clearly thought about how to present their experience for a specific role was telling you something real about their judgment and their interest. The document was never the person, but it correlated with the person.

That correlation has quietly broken. A candidate can now paste a job description into a model and get back a CV that hits every keyword, mirrors the tone of your post, and reads as if they had spent the weekend on it. The same is true of cover letters, take-home assignments, and increasingly the first round of a screening call. The artifacts we built our hiring processes around have become cheap to fake, and cheap to fake means worthless as signal.

I want to be careful here, because this is easy to turn into a complaint about candidates. It isn't. People are using the tools available to them to get past filters that were always a bit arbitrary. The problem isn't dishonesty. The problem is that we are still treating polished output as evidence of underlying capability, when the link between the two has snapped.

What we actually want to know

Strip away the artifacts and the questions underneath are unchanged. Can this person do the work? Will they keep doing it well when it gets hard? Do they fit the way this particular team operates? None of those have ever been answerable from a PDF. We just used the PDF as a convenient first gate and pretended otherwise.

So the honest move is to stop pretending. At RecruiterWings, the company I co-founded, we organize the whole process around one idea: present evidence of capability, not claims of it. That changes what you collect and what you throw away.

Keyword matches go in the bin. What replaces them is harder and more human. A recruiter who actually understands the domain talks to the person about work they have genuinely done, and pushes past the rehearsed version until something specific and unscripted comes out. The detail that can't be generated, because it only exists if you were there, is the signal. How they reasoned through a tradeoff. What they got wrong and what it cost. The part of the problem that turned out to be the real problem.

Verification is the whole job now

The shortlist is no longer a list of people who look right on paper. It's a list of people someone has checked.

This is the shift I keep trying to explain to founders who are hiring. The bottleneck used to be sourcing: finding enough plausible candidates. AI has made the top of the funnel effectively infinite. Anyone can generate a hundred plausible-looking applicants. What is scarce now is verification, and verification doesn't scale the way sourcing does. It requires someone with real judgment spending real attention on each person.

That sounds like a step backward, and in one sense it is. We are returning to a more craft-based, slower kind of assessment. But it is the only thing that survives contact with the new tools. When the cheap signals are gone, you pay for the expensive ones or you hire badly.

The teams that get this right are doing a few concrete things. They use AI aggressively at the top of the funnel, where its job is to widen the net, not to make decisions. They put domain experts, not keyword filters, at the point where capability gets judged. And they measure themselves on whether the people they hire stay and perform, not on how fast they filled the seat. A fast bad hire is just a slow bad hire with extra steps.

The uncomfortable upside

Here is the part I find genuinely encouraging. For years, good people were filtered out by processes that rewarded résumé craft over substance. A brilliant engineer who couldn't be bothered to optimize a CV for an applicant tracking system lost to someone who could. AI has made that particular game unwinnable for everyone, which means we are being forced back toward judging people on what they can actually do.

If you run hiring, the instruction is simple to say and hard to do. Stop trusting the artifact. Build a process where a human who knows the work vouches for each candidate, with reasons you can inspect. It costs more attention. It is also the only thing left that works.