What an AI interview coach actually is
An AI interview coach asks you interview questions, listens to your spoken answer, and scores what it heard against a rubric. Not a chatbot that types questions and reads your typed answer. You speak. It transcribes. It measures clarity, pacing, filler words, answer structure, and how well the content addresses the question. Then it tells you what to change on the next take.
Capstone's coach is the version above this paragraph. Paste a job description, get five questions written for that specific role, and answer them out loud. You get scored written feedback per answer. Runs in the browser, no install, no signup. Sessions stay open for 24 hours so you can come back to the same question set the night before an interview.
What the coach actually scores
Every spoken answer is scored on six dimensions used by workforce programs and structured interview panels:
- Clarity. Is the answer easy to follow. Would a listener retain the point after hearing it once.
- Confidence. Does the delivery carry. Trailing off, hedging language, uptalk on statements.
- Pacing. Words per minute. Interviewers form an impression in the first 15 seconds of an answer; too fast reads as anxious, too slow reads as unprepared.
- Filler words. Density of "um," "uh," "like," "you know." Everyone has a verbal tic under pressure. Recording surfaces yours.
- Answer structure. Did you use STAR (Situation, Task, Action, Result) on a behavioral question. Did you answer the specific question that was asked.
- Content specificity. Adjectives ("passionate," "results-driven") versus evidence ("shipped X in Y weeks"). Interviewers score evidence.
Feedback comes back written per answer, not one overall grade. Specific: what landed, what was vague, what a real interviewer would probe next. You can re-record any answer as many times as you want.
When an AI coach beats a human coach
Three situations where the AI coach is the better tool:
- Reps. The gap between a first take and a fifth take on the same question is enormous. A human coach costs $150 to $300 per session; you get one or two takes. The AI coach runs unlimited retakes at zero incremental cost. Most users sharpen noticeably after 3 to 5 takes on the same question.
- Filler-word and pacing feedback. A human coach can hear that you say "um" a lot. The AI counts them, tells you the exact rate, and tracks it session over session. Structural muscle memory forms faster with a specific number than with a general impression.
- Odd hours + urgency. You have a HireVue link with a Friday deadline. Your coach is booked through Thursday. Availability at midnight is not a coach's job; it is the AI's job.
When a human coach is still worth it
The honest answer: some situations a human coach handles better. Do not pretend otherwise.
- Mid-career pivots with a complex story. A senior engineer moving into product needs someone who has hired product managers to stress-test their positioning. The AI scores the delivery of the story; a human coach helps you decide which story to tell in the first place.
- Deep industry-specific prep. Investment banking superdays, medical residency interviews, tenure-track academic interviews. Formats where the questions themselves require a coach who has been in the room.
- Negotiation strategy. When you have an offer and you are deciding whether to counter, by how much, and how to frame it. This is a decision problem, not a delivery problem.
The right pairing for most people: use the AI coach 5 to 10 times to build the delivery muscle, then spend one session with a human coach on the story and the strategy. The AI is not competing with the human; it is doing the reps a human coach would charge you $600 to do.
Your first week with an AI interview coach
Most people come to a coach when they have an interview 5 to 7 days out. Here is the plan that fits that window:
- Day 1: Baseline record. Paste the job description into the tool above. Answer all 5 questions out loud in one sitting. Do not edit; just notice what happens. Most people discover the middle of their answer collapses around second 40.
- Day 2: STAR the behavioral questions. Rewrite each behavioral answer using the STAR method. Time each answer at 90 seconds spoken.
- Day 3: Fix "tell me about yourself." The opener sets the tone for every question after it. See our TMAY guide. Practice yours out loud three times, timed.
- Day 4: Reps under pressure. Generate a fresh session from the same JD. Answer all 5 questions again, in one sitting, no retakes. This is the real practice; earlier days were setup.
- Day 5: Company + role customization. Use the Role Insights lookup (company + role) to see the traits the target role actually screens for, and get your answers scored against those traits on top of the standard rubric. Record one final full pass.
Total time investment: about 2 hours over 5 days. Compare to the two 45-minute sessions you would get for the same money with a human coach.
How the coach works under the hood
Nothing hidden about the mechanics; they matter for whether the feedback transfers to a real interview.
- Voice, not chat. You speak into the browser microphone. The audio is transcribed and the transcript is what gets scored. Text-based practice teaches you to write, not to speak; interviewers hire on the second.
- Questions from a real JD. The five questions are generated from the responsibilities in the job description you paste in, not from a generic question bank. Practice for the role, not for "interviews in general."
- Same rubric used by structured interview panels. Clarity, confidence, pacing, engagement, persuasiveness, filler-word management. Six dimensions used in workforce programs at NPower, VetsInTech, and the United Nations Foundation.
- Written feedback per answer. Not a single overall grade. Per-answer notes on what landed, what was vague, what an interviewer would probe next.
- Unlimited retakes. The exact thing a real screen does not give you. You practice under time pressure, then run the answer again with the specific fix in mind.
Score your answers against a specific company and role
The default scoring uses a generic rubric. That works for most practice. When you know the specific company and role you are interviewing for, the coach can also score against the traits that role actually screens for.
How it works. On any practice surface (interview, custom interview, live scenario, presentation, and the resume audit), enter a company (optional) and role. The tool generates the 5 to 7 traits that role screens for, plus the skills the role assumes and the red flags to avoid. Every spoken answer then gets a trait-alignment score in addition to the standard rubric. Every resume bullet gets scored against the same traits with a specific rewrite hint for the ones that fall short.
This is what closes the loop between resume prep and interview prep: the same trait profile scores both, so you hear the same feedback in two places. Practical example: if "customer obsession" scores low on your resume bullets, you know to build one specific customer-outcome answer for the behavioral round.
AI interview coach vs. ChatGPT vs. recording on Zoom
ChatGPT. Types questions and reads your typed answer. Skips the entire delivery problem. Interviews are spoken; the practice needs to be spoken too. ChatGPT is a fine tool for brainstorming answers on paper; it is not a coach.
Recording yourself on Zoom. Free and useful. You can hear you were slow or vague. What you cannot do: score yourself against a rubric, generate the right questions for a role you have not interviewed for yet, or track your progress over multiple sessions with consistent measurement. Self-recording is a start; scored practice is what transfers.
Reading interview-questions lists. Zero delivery muscle built. Recognition (I know what to say) is not production (I can say it under pressure). The gap between the two is exactly what breaks candidates in the actual interview.
YouTube walkthroughs. Good for understanding the format of an interview you have never done (federal panels, HireVue, Amazon Loop). Bad for building the muscle of answering. Watch one; then use the coach.
How to get started
Two ways in. Try the free tool above (no signup, one session). If it works for you, the 3-day free trial unlocks unlimited practice, saved session history, per-answer video feedback, tough-conversation scenarios (salary negotiation, difficult manager talks), resume and LinkedIn audits, and the role-and-company customization described above.
Trial pricing after 3 days: $10 per week or $30 per month. Cancel from Settings, one click. We email you the day before the trial ends so there are no surprise charges. The same coaching engine is used inside workforce programs at NPower, VetsInTech, and the United Nations Foundation.