How to Prepare for an AI Interview in 2026: A Recruiter’s 7-Day Plan

Category: Interviews

By AI Career Genie, Career Strategy Team

Your first interview may be with software. What AI interviewers and one-way video screens score, what the recruiter reads afterward, and a 7-day prep plan with a 90-second answer frame, questions to ask, and mistakes to avoid.

*Your first interview for the role may be with software: a recorded one-way screen, or an AI interviewer that asks follow-up questions. Here’s what these systems check, what the people reviewing them see afterward, and a 7-day plan to prepare without sounding scripted.*

> **Quick Verdict:** An AI interview is a structured screen. Often a person reviews the transcript or a summary later. You pass it the same way you pass a good recruiter screen: give clear, specific, job-relevant answers that match your resume. Prepare five or six true stories, practice saying them out loud within a two-minute limit, and don’t use live answer tools or recite scripts. Keyword stuffing doesn’t beat these systems. Consistent, specific evidence does.

Search “how to prepare for an AI interview” and you’ll find the same tips repeated: use STAR, look at the camera, be authentic. Those tips are fine as far as they go. The trouble is that they treat the AI as the audience. After 20 years on the hiring side, we’d put it differently. The AI produces a record, and a recruiter reads that record later in under a minute. You need to prepare for both readers.

---

What an AI interview actually is (three formats)

Built In’s September 2026 explainer describes three formats. Each one rewards slightly different preparation.

| Format | What happens | What to prepare for |

|---|---|---|

| **Asynchronous (one-way) video** | A question appears on screen. You get a short prep window and record an answer, often with a time limit of about two minutes. Some platforms allow retakes. | Answers that stand on their own, a clean start and finish, and timing |

| **Live AI interviewer** | An avatar or voice agent asks questions in real time and can ask follow-up questions based on what you said. | Specific details the AI can follow up on, and staying calm when there’s no human reaction |

| **AI-assisted human interview** | A person interviews you while software transcribes or summarizes the conversation. | Clear, transcribable answers, because the summary is what gets shared |

Glassdoor’s guide to one-way interviews says the prep window is often 30 seconds to two minutes, and it recommends using retakes when the platform offers them. HireVue’s own candidate page says most on-demand interviews give you “just a few minutes” per answer and include a practice question. Do the practice question every time.

How common this is now

These numbers come from vendor surveys, so treat them as directional. They’re still consistent across sources:

  • In the **Greenhouse 2026 Candidate AI Interview Report** (2,950 job seekers in five countries), **63% of US job seekers** said they had already had an AI interview. **70%** said AI wasn’t clearly disclosed beforehand, and one in five found out only when the interview started.
  • The same report found **38%** of US candidates had withdrawn from a process because it included an AI interview. The top trigger was a pre-recorded video scored by AI with no human present. And **51%** of those who completed an AI interview never received an outcome.
  • Greenhouse’s separate *Future of Hiring Candidate Report* found that only 7% of US candidates would refuse an AI interview outright. Of those who had taken one, 70% said it let them show their skills well.

Put simply, the format is common, often sprung on candidates without warning, and not going away. Silence afterward is also common, which is why your follow-up plan matters (more on that below).

---

What the AI is scoring, and what the human reads next

No vendor publishes its scoring model, and criteria vary by employer. Built In’s summary of what these systems generally look at is a reasonable working list:

1. **Content.** It checks clarity (can the transcript capture what you said?), relevance (does the answer connect to the job’s competencies?), structure (a beginning, middle and end), and consistency (do your answers match each other, your resume and your LinkedIn?).

2. **Vocal delivery.** Some systems note pace, pauses and filler words.

3. **Body language.** Some systems look at eye contact, expression and posture. This is the most contested category. HireVue stopped using facial analysis in 2021 after a complaint was filed with the FTC.

Now the hiring-side part. Whatever the score, the next reader is usually a recruiter or hiring manager. That person opens a scorecard, a transcript or a highlight clip, and asks the same three questions they ask in any phone screen:

  • **Can this person do the job we posted?** That means evidence tied to the job description.
  • **Is the story true and consistent?** The resume, LinkedIn and answers should all agree.
  • **Would I spend 30 minutes talking to them?** You come across as clear, direct and human.

That’s why “talk to the algorithm” advice backfires. Repeating keywords makes a transcript worse to read. A specific number, a decision you made, and a result you can explain make it better for both readers.

> Match-score theater doesn’t work in a transcript either. One true story with a number beats five buzzwords.

---

Where the advice conflicts (and what we’d do)

The top results disagree on a couple of points, so here they are:

  • **Eye contact.** Built In reports that many platforms use eye tracking to flag script-reading or real-time AI assistance. HireVue’s own candidate page says “don’t worry about making eye contact with the camera.” *Our take:* look at the lens most of the time. It reads as confident on a recording a human will watch. Don’t perform it, though.
  • **Notes.** HireVue says, “You can even keep your notes next to you.” Other guides warn against anything that looks like reading. *Our take:* keep a single index card with three to five words per story. Use it to jog your memory, not as a script. If you’re reading full sentences, you’ll sound like you’re reading.
  • **Keywords.** Built In says grounding answers in the job description’s language helps, while its executive-recruiter source warns against trying to “game the algorithm” by stuffing keywords. *Our take:* use the posting’s exact names for the tools and skills you actually used, once each, inside a story. Never as a list.

When the advice conflicts, follow the employer’s instructions and the platform’s practice screen. Those are the only rules that apply to your interview.

---

The 7-day AI interview prep plan

You often get the invitation with a deadline only a few days away. Here’s how to use a week, or compress it into three days.

Day 1: Decode the posting

  • Open the job description and highlight the five to seven competencies it repeats, such as stakeholder management, SQL, forecasting or customer escalations.
  • Write each one as a likely question: “Tell me about a time you…”
  • Check the invitation for format, time limits, retakes, and whether a person reviews the results. If it doesn’t say, ask (see the questions below).

Day 2: Build five or six stories you can tell without notes

Built In’s experts recommend five or six stories covering a big accomplishment, a failure, a conflict, a difficult decision, and something you built or changed. For each one, write down:

  • One sentence of context
  • What **you** did (use “I,” not “we”)
  • One number you can source
  • What you’d do differently

Map each story to two or three competencies from Day 1. Six good stories can cover most behavioral questions. Our [STAR method guide](/blog/star-method-interview-questions-2026/) includes a scoring rubric if you want to grade your own answers.

Day 3: Make sure your resume and answers tell the same story

Consistency is one of the signals these systems check, and it’s definitely what the human reviewer checks. Read your resume next to your stories. Titles, dates, team sizes and metrics should match exactly. If a bullet says “led” and your story says you “helped,” fix one of them. If you used AI on your resume, now is the time to remove anything you can’t talk about for two minutes. Our guide on [humanizing AI-written resume copy](/blog/can-recruiters-spot-chatgpt-resume-humanize-ai-writing/) covers what to look for.

Not sure your resume even matches this posting? Run it through the [free resume scorer](/free-resume-scorer/) with the job description. It’s free with no signup, and you get a match score, a keyword gap count and one sample bullet rewrite. Close the gaps that are real. Don’t add skills you can’t defend.

Day 4: Practice out loud, on a timer

The AI interview advice everyone agrees on: **interviewing is a speaking skill.** Record yourself on your phone answering three questions in under two minutes each. Then read the transcript (your phone’s voice memo transcription works). Look for:

  • Words the transcription got wrong. Names, acronyms and jargon often get mangled, so say them clearly or add a plain-English phrase.
  • Openings that don’t answer the question. If your first sentence doesn’t, start over.
  • Filler and backtracking. Cut “so basically” and “kind of.”

Day 5: Rehearse with an interviewer that asks follow-ups

Live AI interviewers ask follow-up questions. Greenhouse’s description of a good AI screen is one that asks for “one specific number, one decision they made or one thing that didn’t go to plan.” Practice with a tool that does the same. Our [AI mock interviews built from the real job description](/blog/ai-mock-interviews-from-the-job-description/) generate questions from the posting you paste in and push for specifics. Any mock tool that asks follow-ups will help. The point is to get used to answering with no nods or smiles to guide you. If that silence throws you, our [interview anxiety guide](/blog/overcoming-interview-anxiety-ai-mock-interview/) has some drills.

Day 6: Set up and do a dress rehearsal

  • Light in front of you, camera at eye level, quiet room, laptop plugged in, notifications off.
  • Do the platform’s practice question, and do it twice if you can.
  • Put one index card below the camera with your story keywords on it.
  • Decide your 30-second opener for “Tell me about yourself”: current role, one proof point, and why this role.

Day 7: Take it, then follow up

  • Use the full prep window to jot down two or three points before recording.
  • If retakes are allowed, use at most one per question. A second take is usually better. A fifth usually sounds rehearsed.
  • Within 24 hours, email the recruiter (not the platform). Confirm you completed it, add one sentence connecting a story to the role, and ask about the timeline. With half of AI interviewees never hearing back, according to Greenhouse, a short follow-up to a real person matters. Our [follow-up email templates](/blog/follow-up-email-after-interview-no-response-2026/) give you the wording and the timing.

---

A 90-second answer frame that works for both readers

Use this for nearly any behavioral question:

1. **Headline (one sentence).** Answer the question directly. CNBC’s September 2026 coverage quotes a hiring-tech expert recommending you open with a “very, very clear yes or no” and warning that vague answers are the most common mistake. “Yes. The clearest example is when I rebuilt our onboarding process for enterprise accounts.”

2. **Context (one or two sentences).** Scope and stakes, with one number.

3. **What I did (three or four sentences).** Your decisions, in order, using the job description’s tool or skill names where they’re true.

4. **Result (one sentence).** A number you can source, or a specific change.

5. **Close (one sentence).** What you learned, or why it matters for this role.

Before and after

These are illustrative examples, not a real candidate.

**Before (sounds scripted, reads vague):**

> “I’m a results-driven professional who leverages cross-functional collaboration to drive stakeholder alignment. In my previous role, I spearheaded many initiatives that improved efficiency and exceeded expectations.”

**After (transcribable, specific, defensible):**

> “The best example is our customer onboarding rebuild last year. We had 14 enterprise accounts waiting six weeks to go live. I mapped every handoff, cut two approval steps, and built a shared checklist in Asana that sales and support both used. Average go-live dropped to about three weeks over the next two quarters. What I took from it: most delays were handoff problems, not people problems, and that’s the kind of work this role describes.”

The second answer gives a live AI something to follow up on (“what broke along the way?”), and it gives the human reviewer a reason to schedule a call.

---

Questions you’re allowed to ask before the interview

Candidates often don’t realize they can ask. In Greenhouse’s data, 39% of US candidates wanted an explanation of what the AI measures, 38% wanted confirmation that a person reviews the results, and 46% wanted the option to request a human interview. A short, polite email to the recruiter is normal:

  • “Will a person review my responses, or is this an automated decision?”
  • “What competencies is the interview evaluating?”
  • “Are retakes allowed, and is there a time limit per answer?”
  • “Is an alternative format available if I need one?”

**Accommodations.** If a disability affects your speech, timing, camera use or processing, ask for a reasonable accommodation before you start, such as extra time, a written format, or a live interviewer. The ADA’s accommodation duty applies to the hiring process. Put the request in writing.

**State rules vary.** Built In notes that Illinois requires employers to get consent and explain how an AI system will evaluate video interviews, that Maryland requires consent before facial analysis is used, and that New York City and Colorado have passed rules on AI in hiring. This is general information, not legal advice. If you’re concerned about a specific decision, talk to a qualified employment-law resource.

---

What not to do

  • **Don’t use a live AI answer tool.** Built In reports that platforms watch for real-time assistance. Even when they don’t catch it, a human hears the pause-and-recite rhythm. If you’re found out, you lose credibility, and almost certainly the candidacy too.
  • **Don’t recite a script you had AI write.** Harvard’s career center, Built In’s experts and HireVue all say the same thing: prepare points, not paragraphs.
  • **Don’t stuff keywords.** Ten tools in one answer reads as a list, not experience.
  • **Don’t claim skills you can’t back up.** A live AI follow-up, or the human round after it, is exactly where padded skills fall apart.
  • **Don’t skip the practice question.** It’s free information about the platform.
  • **Don’t assume silence means rejection.** It often means no one has looked yet. Follow up with a person.

---

How this fits a 7–14 day interview push

An AI screen is a gate in the middle of the funnel. If you aren’t getting invited to them at all, the problem is earlier, and our guide to [why you’re not getting interviews](/blog/why-am-i-not-getting-interviews-2026/) helps diagnose it. If you’re getting screens but not second rounds, it’s usually story quality and consistency. A realistic two-week plan:

  • **Days 1–3:** Tighten your resume against two or three target postings, make your LinkedIn match it ([LinkedIn optimization guide](/blog/linkedin-profile-optimization-2026/)), and build your five or six stories.
  • **Days 4–7:** Apply to roles that genuinely fit and practice two mock sessions with follow-up questions.
  • **Days 7–14:** Take screens as they arrive, follow up within 24 hours, and send [thank-you notes](/blog/thank-you-email-after-interview-templates-2026/) after any human round.

No tool can promise you an interview, and we won’t. But consistent materials plus rehearsed, specific stories are what we see move candidates from “screened” to “scheduled.”

---

FAQ

What is an AI interview?

An AI interview is a screening interview run by software instead of a person, or alongside one. It can be a recorded one-way video, a live AI interviewer that asks follow-up questions, or a human interview that AI transcribes and summarizes. The system evaluates your answers, and a recruiter usually reviews the results.

How do I pass an AI interview?

Prepare five or six true stories mapped to the job description. Answer each question directly in under about two minutes using a clear structure (headline, context, what you did, result). Speak clearly enough to be transcribed, and keep your answers consistent with your resume. Practice out loud with follow-up questions beforehand.

Can AI interviews detect reading from notes or using ChatGPT?

Some platforms say they monitor for script-reading or real-time assistance, according to Built In’s reporting, though capabilities vary and aren’t published. HireVue says brief notes are fine. A short keyword card is low-risk. Reading full answers or using a live answer tool isn’t worth it.

Does eye contact matter in an AI interview?

The guidance conflicts. Some sources say systems analyze it, while HireVue tells candidates not to worry about it. Looking at the lens most of the time reads well to the human who watches the recording later, so it’s a sensible default.

Do I have to agree to an AI interview?

You can decline, but it usually means withdrawing from that process. You can also ask whether a human reviews the results, what’s being measured, and whether an alternative format or an accommodation is available. Some states, including Illinois, require notice and consent for AI evaluation of video interviews.

How long until I hear back after an AI interview?

It varies. Greenhouse found that 51% of US candidates who completed an AI interview never received an outcome. Send a short follow-up to the recruiter within 24 hours and again after about a week.

---

Soft CTA: Prepare for the screen, not just the score

1. **Check the gap.** Paste the job description into the [free resume scorer](/free-resume-scorer/). It’s free with no signup.

2. **Fix what’s real.** Rewrite four to six bullets so they match the stories you’ll tell.

3. **Rehearse with follow-ups.** Practice two [mock interviews from the job description](/blog/ai-mock-interviews-from-the-job-description/).

4. **Getting ghosted after screens?** Get a [free Resume Ghosting Report](/ghosting-report/). It takes 60 seconds, needs no credit card, and was built by recruiters for candidates.

5. **Want unlimited rewrites, a LinkedIn optimizer and JD-based mock interviews in one place?** See [pricing](/pricing/): $29 a week, or $99 a month with one 60-minute coaching session included. Cancel anytime, with a 48-hour money-back guarantee.

Related reads: [STAR method interview questions](/blog/star-method-interview-questions-2026/) · [Follow-up email after no response](/blog/follow-up-email-after-interview-no-response-2026/) · [Can recruiters spot a ChatGPT resume?](/blog/can-recruiters-spot-chatgpt-resume-humanize-ai-writing/)

*Sources: [Built In, “Your Next Job Interview Could Be With a Bot”](https://builtin.com/articles/how-to-prepare-for-an-ai-interview) (Sept 16, 2026); [Greenhouse 2026 Candidate AI Interview Report](https://www.greenhouse.com/blog/2026-candidate-ai-interview-report) (Apr 29, 2026); Greenhouse 2026 Future of Hiring Candidate Report; HireVue candidate interview tips; Glassdoor video interview guide; [CNBC Make It, “Your AI interviewer will see you now”](https://www.cnbc.com/make-it/2026/09/16/your-ai-interviewer-will-see-you-now-what-job-seekers-should-know-about-new-hiring-tech.html) (Sept 16, 2026); Harvard FAS career services guidance on using AI to prepare for interviews. Survey figures are as reported by those organizations.*

[Back to all articles](/blog/) | [Try the free resume scorer](/free-resume-scorer/) | [View pricing](/pricing/)