What to Screen On When Every Resume Is AI-Polished

A resume used to tell you something even when it was badly written. What a person chose to put first, how they described a job, which skills they bothered to mention: those were their own decisions, and decisions carry information. A recruiter skimming a pile of resumes was partly reading the jobs and partly reading the choices.

AI writing tools have mostly removed the choices. Ask a model to tailor a resume to a job description and it will pick the keywords, framing, and order it thinks your ATS wants. When everyone does that, resumes start to look alike, and a filter that ranks them is mostly ranking how well each candidate's AI tool guessed at your scoring criteria. This isn't a fringe case, either. One 2024 survey found that 46% of job seekers had used AI tools to write or substantially revise their resumes.

The same tools made applying cheap, so you also get more of these resumes. iCIMS data from 2025 shows applications per open role rose 40% year over year, concentrated in customer service, sales, and operations roles. Those are the high-volume jobs where recruiting teams were already stretched. There are more applications but not more qualified candidates, so the pile grew while the number of people in it worth hiring stayed about the same.

The resume was never a strong filter

I'd go further, though. Even before any of this, the resume wasn't much of a filter. The things you can read off one, like years of experience, have long been found to predict job performance much less well than a structured interview does. People reading resumes were already making a lot of their calls on weak evidence. They just did it at lower volume and without noticing much.

The teams that saw this early moved the real qualification step later, into structured phone screens, work samples, or behavioral assessments. That helped. The catch is that structured phone screens don't scale to 300 applicants. A recruiter can do perhaps eight to twelve good screens in a day, so a high-volume role gets either a rushed screen or a queue that runs for weeks. Neither gives you what you want, which is a consistent, careful look at everyone in the pool, done quickly.

That combination is what's breaking a lot of pipelines: volume at the top inflated by AI, and a limit in the middle on how many conversations people can have. A smarter resume filter doesn't help with either, since filters are exactly what AI-written applications are designed to beat.

Moving the conversation to the front

What you'd want, then, is the structured conversation, the part that actually predicts something, placed where the resume used to be and given to every applicant. That's the case for voice screening. A live conversation can't be generated the way a document can. In a ten-minute voice interview you hear how someone handles a hard question, how specific they get about a project they really worked on compared with one that was padded onto the resume, and what they do with a follow-up they didn't expect. Those things happen while people talk, and they're hard to rehearse convincingly.

Asendia is our attempt to make that practical. Instead of a filter scoring documents that were written to beat it, every applicant gets that structured conversation with our software, however many applicants there are, so nobody waits days for a recruiter to open the queue, and a weekend that brings in 400 applications can be worked through before Monday. What comes out is a ranked shortlist with structured notes on what each candidate said, what stood out, and why they ranked where they did, and it goes into the ATS for recruiters to pick up. For more on the difference between AI that assists and AI that moves a pipeline forward on its own, see our post on agentic recruiting.

I don't think the AI resume problem fixes itself. The tools are too good and too easy to use, and they're aimed squarely at keyword filters. Improving your resume screening now means getting better at something that has stopped telling you much. The better move, I think, is to put the conversation as early in the process as you can and to run it the same way for everyone.

Ready to transform your hiring strategy? Schedule a Demo with our founders today!