How to tell if a recruiter is AI

Three different things get called an AI recruiter, and they call for different responses.

• Automated sourcing. Outreach assembled from a template and sent by software, with a real person's name in the signature.
• An AI screener. A voice agent or chat bot that runs the first round, asks a fixed list of questions, and hands a transcript and a score to the hiring team.
• A person using AI tools. A recruiter who reads your reply but relies on resume parsing, ranking, and a notetaker to get through the pile.

The third case is the most common and the hardest to see from the outside. What most people are really asking is whether a human will read what they wrote, which is a question about the stage, not about the sender.

What each kind of screening tool can and cannot see is covered across the detectability cluster.

You would rather know before you reply than after. The checks below are the ones that hold up, and they come down to three things: timing, specificity, and what happens when you ask something off-script.

• Replies land in seconds, at any hour, including weekends.
• The message names your job title and city but nothing from your actual work.
• You ask something specific, such as team size or why the last person left, and get back a reply that answers a different question or repeats the pitch.
• The signature name has no posting history, or the profile was created weeks ago.
• Every path leads to a scheduling link rather than a conversation.

The most reliable test is a question that needs judgment: ask which part of your background made them reach out. A person names something concrete. A template restates the job.

None of this is proof. Recruiters working through hundreds of profiles also send templates and answer in one line, and some AI outreach now waits before replying precisely because instant answers gave it away.

Voice agents have a rhythm that is hard to hide once you know it.

• It never interrupts, and it begins the next question a beat after you stop talking, every time.
• Pacing and tone stay flat whether you laugh, apologize, or pause to think.
• Asking it to repeat only the second half of a question produces the whole prompt again, word for word.
• It ignores anything you volunteer that is not an answer to the current question.
• There is no small talk at either end, and no real answer to a question about the team.

A data analyst interviewing for a mid-level role at a grocery chain joined what the invite called a live screening. The interviewer asked five questions in a fixed order, never picked up on the pipeline work she described, and closed at exactly twenty minutes. She decided to treat the round as recorded and structured, which was the right call whether or not a person was listening.

Rehearsing that format is what the mock interview page is for: an AI interviewer asks and scores, so the rhythm is familiar before it counts.

The preparation barely changes once you accept you may never know for certain.

• Say your name and the role at the start, since automated transcripts attribute speakers better with a clean opening.
• Keep every answer self-contained. Do not lean on something you said three questions ago.
• Use the job description's own words where they are honestly true of you, because the transcript may be matched against it.
• Ask in writing whether a human reviews the recording, and at which stage you speak to one.

Where SubcueAI fits, stated plainly: it listens to a live conversation and drafts an answer on your screen. When the call runs in a browser tab, the extension side panel captures the meeting tab's audio, which is the interviewer and not your microphone. For a desktop meeting client such as Zoom, Google Meet, or Microsoft Teams, the macOS and Windows app captures system audio and your mic. No bot joins the call, and nothing is injected into the meeting page. A recorded one-way video, a proctored environment, a company-managed laptop, or a shared screen sits outside that, and there the honest answer is preparation rather than assistance.

Setup for each surface is walked through on the tutorial page.