Skip to content

AI copilot for interviewers · Google Meet

Focus on the candidate. Let AI keep track of the interview.

A live copilot for Google Meet interviews. Speaker-labelled transcript, next-question suggestions as the candidate speaks, and a finished scorecard when you hang up — with no bot in the call and no audio recorded.

Add to Chrome — it’s free See it in action
Works with Google Meet Zoom soon Microsoft Teams soon No credit card required
meet.google.com/abc-defg-hij
Live
Transcript
Senior Backend · Round 2
    Copilot
    Focus coverage
    System design
    Ownership
    Communication
    Steer the copilot: “go deeper on incidents”
    0
    bots in your call

    Runs in the interviewer’s browser — nothing joins the meeting

    2,000+
    interviewers on Chrome

    Installed straight from the Chrome Web Store

    50,000+
    interviews analysed

    Transcribed live and turned into a scorecard

    4.8
    on the Chrome Web Store

    Based on 120+ interviewer reviews

    Real-time analysis

    The candidate is still talking. The copilot has already caught it.

    Every finished sentence runs through the same four-step loop while the answer is still going: what was claimed, what backs it up, which focus area it belongs to, and what you should ask next.

    • Claims without proof

      “We improved performance a lot.” The copilot asks for the number so you don’t have to remember to.

    • “We” hiding “I”

      Team outcomes that never reveal the candidate’s own decision get surfaced as worth probing.

    • Focus areas still empty

      Fifteen minutes left and nothing on mentoring yet? You see it during the call, not in the debrief.

    Product

    Everything an interviewer needs, from “hello” to hired.

    Live transcript, real-time guidance and evidence-backed scorecards — in one panel next to the call.

    Live transcript

    See every answer as it’s said — labelled by speaker.

    The copilot reads Google Meet’s own live captions, so you get an accurate, speaker-labelled transcript in real time without recording audio or inviting a bot.

    • No bot, no recording

      Nothing joins the call. Nothing is visible to the candidate. Audio never leaves Meet.

    • You vs. candidate, automatically

      Your own lines are detected and separated, so notes and analysis only ever focus on the candidate.

    • 60+ languages

      Interview in the candidate’s language. Switch the transcript language mid-call with one click.

    Live · 42 sentences EN → switch
    • You

      Walk me through how you’d design the ingestion pipeline.

    • Candidate

      I’d start from the write path: an append-only log in front, consumers that are idempotent so retries are safe…

    • Candidate

      …and I’d keep the schema registry outside the hot path so producers never block on it.

    • You

      What would you monitor first?

    No audio Hidden from candidate 60+ languages

    In-interview AI

    Always know what to ask next.

    As the candidate answers, the copilot proposes follow-ups, flags vague claims, highlights strong signals and tracks which of your focus areas are still uncovered.

    • Follow-up & dig-deeper prompts

      Concrete next questions grounded in what was just said — not generic question banks.

    • Signals and red flags

      “Strong signal”, “watch out” and “missing area” cards so you don’t lose the thread while listening.

    • Steer it in plain language

      Type “go technical”, “shorter questions” or “candidate may be using AI” and the copilot adapts for the rest of the call.

    Follow-up ⧉ copy

    What did you stop doing to make time for the migration?

    Watch out ⧉ copy

    Answer is about the team’s outcome — ask for their personal decision.

    Strong signal ⧉ copy

    Brought up trade-offs unprompted (consistency vs. latency).

    Missing area ⧉ copy

    Nothing yet on mentoring — planned focus area.

    candidate may be using AI — make the questions harder applied

    After the call

    Walk out with the scorecard already written.

    Your rubric, filled in with evidence. Each rating links back to the exact quote from the transcript, so debriefs are about the candidate, not about who remembers what.

    • Structured scorecards

      Ratings per competency, strengths, risks and suggested next-round questions.

    • Evidence, not vibes

      Every claim in the summary cites the transcript line it came from.

    • Fit summary against the job

      Compare the conversation with the job description and the CV to spot gaps before the debrief.

    Scorecard
    Senior Backend Engineer · Round 2
    Ready
    • System design

      [12:41] Idempotent consumers, schema registry off hot path.

    • Ownership

      [12:41] “I paged on-call together with my lead.”

    • Communication

      [12:41] Clear, occasionally long-winded under pressure.

    • Mentoring

      [12:41] Not covered — suggested for round 3.

    PDF Markdown JSON → ATS

    For hiring teams

    Consistent interviews across every panel.

    Define focus areas, must-ask questions and the scoring rubric once per role. Every interviewer gets the same kit, the same prompts and produces a comparable scorecard.

    • Role kits

      Focus areas, competencies and pre-brief context (CV, JD, take-home) per position.

    • Calibrated scoring

      Same rubric, same evidence format — so panels can actually be compared.

    • Exports & webhooks

      Push scorecards to your ATS or knowledge base as Markdown, PDF or JSON.

    Senior Backend

    rubric v3 · 6 interviewers

    • System design
    • Ownership
    • Communication
    • Mentoring
    Account Executive

    rubric v1 · 3 interviewers

    • Discovery
    • Objection handling
    • Pipeline rigor
    Panel comparison
    3 of 3 scorecards in · 1 disagreement on Ownership
    Debrief ready

    AI on your transcript

    Ask your interview anything — mid-call or a month later.

    The transcript isn’t a wall of text you never open again. It’s something you can question, summarise and turn into work with a single click.

    Ask AI custom questions

    Ask in plain language and get an answer grounded in the transcript, with the exact line and timestamp it came from.

    Did they ever give a concrete number?

    AI

    Three times: 10× launch traffic, p95 from 800 ms to 120 ms, and a nine-minute rollback. The 40% cost claim is still unquantified.

    Grounded in 3 transcript lines
    • 00:12:41 We hit 10× traffic during the launch week.
    • 00:19:03 p95 went from 800 ms down to about 120 ms.
    • Did they ever give a concrete number?
    • What haven’t I covered yet?
    • Where does this answer contradict the CV?
    • Summarise the last ten minutes for my colleague.
    updating

    Summarised as it happens

    A running summary that rewrites itself every few minutes — so a colleague joining at minute 20 catches up in ten seconds.

    • Scaled a launch to 10× traffic; owned the read-replica migration.
    • Cost claim (40%) still has no baseline.

    Action items, written for you

    Next-round questions, take-home ideas, reference checks and the follow-up email, generated from what was actually said.

    • Send the take-home: cost-aware batch scheduling
    • Round 3 must-ask: mentoring evidence
    • Reference check: who owned the migration?
    Add all to the debrief Send to ATS

    Previous rounds, right there

    Earlier interviews with the same candidate summarised beside the call, so the panel never asks the same question twice.

    • Round 1 · Recruiter screen

      Strong on scale stories, vague on numbers.

    • Round 2 · Take-home review

      Clean tests; skipped the rollback question.

    Actionable insights, one click each

    The same prompts your team runs after every interview, as buttons instead of copy-paste.

    • Summarise the interview
    • Red flags & inconsistencies
    • Compare against the job description
    • Coverage gaps for round 3
    • Draft the debrief note
    • Turn into a scorecard
    • Questions for the reference check
    • Rewrite as candidate feedback

    How it works

    Up and running before your next interview.

    No IT ticket, no calendar integration, no bot to invite. Install the extension and join the call.

    1. 01
      Install and pick the role

      Add the Chrome extension, choose or create a role kit: focus areas, rubric and optional context such as the CV or job description.

    2. 02
      Join the Google Meet

      The copilot panel opens beside the call. No bot, no extra participant — the candidate experience is unchanged.

    3. 03
      Interview with a second brain

      Read the live transcript, take the suggested follow-ups you like, steer the copilot in plain language.

    4. 04
      Hang up with the scorecard done

      Review the evidence-backed scorecard, adjust, and share it with the panel or push it to your ATS.

    Compare

    A copilot, not another notetaker bot.

    Meeting bots summarise after the fact. An interview copilot helps while the decision is still being made — and never shows up in the candidate’s participant list.

    Capability Kettle Notes Notetaker bots
    Joins the call as a participant
    Visible to the candidate
    Records or processes audio
    Helps during the interview
    Running summary while you interview
    Ask the transcript questions mid-call
    Speaker-labelled live transcript
    Scorecard aligned to your rubric
    Tracks focus-area coverage live
    Bring your own model key

    Use cases

    Built for every team that interviews.

    Recruiters get consistency. Hiring managers get confidence. Candidates get a better-run interview.

    • Engineering hiring

      System design and behavioural rounds with follow-ups that actually probe depth.

    • Sales & customer teams

      Spot coached answers, check for numbers, and keep discovery-style interviews on track.

    • High-volume recruiting

      Same questions, same rubric, comparable scorecards across hundreds of screens.

    • Hiring managers

      For people who interview twice a quarter and shouldn’t need to be interview experts.

    • Agencies & RPOs

      Client-ready, evidence-backed write-ups minutes after the call ends.

    • Panel interviews

      Each interviewer sees the coverage map, so panels stop asking the same question twice.

    Security & privacy

    Built to be safe to use in front of candidates.

    Privacy is a product requirement, not a checkbox. The copilot is designed so the interview looks and sounds exactly like it would without it.

    • GDPR-ready
    • DPA available
    • EU or US data region
    • SSO (SAML) on Enterprise

    • No audio, no recording

      We read Meet’s live captions. We never access your microphone or record the meeting.

    • Invisible to candidates

      No bot, no extra participant, no caption box on the candidate’s screen.

    • Encrypted in transit and at rest

      TLS everywhere, encrypted storage, per-workspace data isolation.

    • You own the data

      Configurable retention, one-click delete, export anything at any time.

    • Bring your own model key

      Route AI calls through your own provider account so transcripts stay within your agreements.

    • Human decides, always

      The copilot suggests; it never scores a candidate automatically or makes a hiring decision.

    Pricing

    Simple pricing for individuals and teams.

    Start free. Upgrade when the whole panel wants in.

    Free
    $0

    For individual interviewers getting started.

    Add to Chrome
    • 5 interviews / month
    • Live speaker-labelled transcript
    • Next-question suggestions
    • 1 role kit
    • Markdown export
    Most popular
    Team
    $24 per interviewer / month

    For hiring teams that want consistent, comparable interviews.

    Start 14-day trial
    • Unlimited interviews
    • Evidence-backed scorecards
    • Shared role kits & rubrics
    • CV + JD pre-brief
    • Webhooks, PDF & JSON export
    • Admin controls & usage reports
    Enterprise
    Custom

    For organisations with security and procurement requirements.

    Talk to sales
    • Everything in Team
    • SSO (SAML) & SCIM
    • EU or US data residency
    • Bring your own model key
    • DPA & security review
    • Priority support & onboarding

    Prices in USD. Annual plans billed yearly. Taxes may apply.

    FAQ

    Questions interviewers ask us.

    Something else? Write to us — a human answers.

    Does a bot join my interviews?

    No. Kettle Notes is a Chrome extension that runs inside the interviewer’s own Meet tab. There is no extra participant, nothing appears in the participant list, and the candidate sees a completely normal call.

    Do you record audio or video?

    No. The copilot reads the live caption stream that Google Meet already produces for the meeting. We never request microphone or camera permissions and nothing is recorded.

    Can the candidate tell that I am using it?

    No. The panel lives in your browser only. The caption box that Meet would normally show is hidden on your side and was never visible on theirs.

    Which meeting platforms are supported?

    Google Meet today. Zoom and Microsoft Teams are on the roadmap. Join the waitlist from the pricing section if you need them first.

    Which languages does the transcript support?

    Everything Google Meet captions support — currently 60+ languages. You can switch the transcript language during the call and the AI suggestions follow the language of the conversation.

    What happens to my interview data?

    Transcripts and scorecards are stored encrypted per workspace, with a retention period you control and one-click deletion. Enterprise customers can choose an EU or US region and route AI calls through their own model provider key.

    Does it integrate with my ATS?

    Scorecards export as Markdown, PDF or JSON and can be pushed anywhere with webhooks. Native integrations with popular ATS tools are in progress — tell us which one you use.

    Is it fair to candidates?

    The copilot supports structured interviewing: the same focus areas and rubric for everyone, with evidence tied to what was actually said. It suggests questions and organises evidence; the interviewer makes every judgement. It never auto-scores or auto-rejects a candidate.

    Your next interview, better prepared.

    Install in a minute. Use it on your very next Google Meet.

    Add to Chrome — it’s free Book a 20-minute demo