Hiring signal
How do you find great candidates when every resume is AI-perfect?
Treat the resume as a starting point, not a verdict. The reliable signal is what a candidate can explain, prioritize, and produce when the prompt resembles the job they want.
01Why polished resumes are a weaker filter now
AI can help a candidate find missing keywords, tighten a summary, and turn uneven experience into a cleaner story. That does not automatically make the candidate unqualified. It does mean formatting, fluency, and perfect alignment are less useful as proof of ability than they used to be.
A resume still answers important questions: what work has this person done, what context have they operated in, and what should we explore? It is simply a poor final answer to whether they can handle the decisions in this role.
02Replace keyword confidence with job-shaped evidence
Choose three to five traits that matter for the open role, then ask every serious applicant the same short questions about realistic work. A support candidate might respond to a failed rollout; a product candidate might choose between a custom feature and reliability work; an engineer might explain their first hour after an incident.
Good prompts require prioritization, tradeoffs, and a sequence of actions. They are difficult to answer convincingly with generic language because the candidate has to make the work concrete. Keep the exercise short, accessible, and relevant to the job rather than asking for unpaid production work.
03Use AI as a drafting assistant, not the decision-maker
AI can turn a job description into a first draft of traits, questions, and a scorecard. A recruiter or hiring manager should edit that draft, read the full response, and own the decision. The system should help the team become more consistent, not conceal why someone moved forward or was declined.
In Honrly, the assessment is attached to the applicant record: reviewers can see the question, the score, a supporting quote, and the whole answer. That makes a number discussable instead of magical.
04Screen for signal before you schedule interviews
A practical first pass is resume plus a short work sample. A second reviewer checks the same evidence using the same rubric. Only then does the team spend a live interview testing the areas that remain uncertain. This preserves scarce calendar time and gives interviewers a shared starting point.
For roles where speaking is central to the work, add a brief asynchronous video response. For roles where clear writing matters, keep the assessment written. The point is not to make every person perform in the same medium; it is to assess the job fairly.
05Measure the process, not just the resume
Audit which applicants advance, how often reviewers agree, time to review, candidate completion, and later on-the-job outcomes. Look for steps that add delay without changing a decision. A strong hiring process gets more useful evidence with less burden on candidates and the team.
This is a useful lens whether you are considering a large recruiting suite such as Ashby or Eightfold, or a focused hiring workflow. The durable advantage is not another resume-ranking model. It is a review process that connects the job, evidence, and human judgment.
Questions
Should we reject candidates for using AI on their resumes?
No. AI assistance may be normal for writing. Evaluate the evidence relevant to the role: a short, job-specific response, a consistent rubric, and a conversation that probes the candidate's thinking.
What is the best way to evaluate candidates beyond a resume?
Use a brief job-relevant work sample or structured assessment, have reviewers use the same criteria, and use interviews to investigate the evidence rather than start from scratch.
How long should a pre-interview assessment be?
Keep it proportional to the role and stage. A few focused questions that can be completed in a reasonable amount of time are more respectful and often more revealing than a long take-home assignment.