NAX Ai Interviewer replaces the entire first round. Here's exactly what happens between sending an invite and reading a scored result.
A short slide-based explainer put together by our team: how a NAX Ai Interviewer session is set up, what the candidate goes through, and what lands on your desk afterwards.
An illustrated guide video — slides and narration only, no actors and no filmed footage.
HR generates an invite link per candidate and role. Links are one-time and expire on your schedule — no duplicate attempts, no shared back-doors.
Details, resume upload and the role being applied for. Everything is captured before a single second of interview time is spent.
Camera, microphone, speaker, lighting and network are tested and scored before recording starts. Bad setups get fixed, not discovered later.
The AI interviewer asks role-specific questions out loud and listens. It follows up on vague answers, switches between BM and English, and gives 30 seconds of think time per question.
One continuous take from the first question to the last, while anti-cheat watches for tab switches, extra monitors, copy-paste, multiple faces and long silence.
Answers are scored on relevance, clarity and depth. HR gets a Pass / Review / Fail decision with transcript, recording and the full integrity log attached.
One take, compressed and stored privately. Jump straight to any question with transcript timestamps.
Pass / Review / Fail with a per-question breakdown on relevance, clarity and depth.
Every anti-cheat event with a timestamp, so a borderline result can be judged in context.
Resume, contact details and reference code, all attached to the same record.
A NAX interview runs about 18 minutes end to end. Here is the whole thing, in order, from the applicant's side of the screen — no jargon, just what they see and hear.
No app to install and no account to create. The invite opens in whatever browser the candidate already has, on a phone or a laptop, and the page tells them in plain language how long it will take and what will be recorded.
Name, contact, the role they applied for, and their resume. It takes about ninety seconds. The moment they submit it, they get a reference code they can quote later if they want to ask about their application.
Camera, microphone, speaker, lighting and connection are all tested while they watch. If the room is too dark or the mic is picking up a fan, they are told before recording starts — not after they have given a good answer nobody can hear.
The AI greets them by name in the language they chose, explains the rules once, and asks the opening question out loud. There is a short think-time countdown, then the microphone opens.
They answer in their own words. If an answer is vague, the AI asks the obvious follow-up a good interviewer would ask. If they ramble, it waits. If they freeze, it offers to repeat the question. Nothing about it feels like filling in a form.
Technical roles get a short live task instead of a final question. Then the AI thanks them, tells them what happens next and when to expect an answer, and stops the recording in one clean take.
Within seconds the interview appears in your review queue with a score, a transcript, the recording and the integrity log. That is the first moment a human is needed at all.
Plenty of hiring tools overpromise. Here is where we draw the line on purpose, so nobody is surprised three months in.
NAX never sends a rejection on its own and never picks your new employee. It sorts a pile of applicants into Pass, Review and Fail so your team spends its afternoon on the top of the pile instead of the whole thing.
There is no scoring of appearance, accent, expression or 'confidence'. The camera exists so you can see the same person answered the whole way through, and nothing else.
A single tab switch is a note in the log, not a verdict. Only a pattern of behaviour ends a session early, and even then the recording is kept so a human can look and disagree.
The deep conversation about salary, culture and the messy specifics of the job stays with your team. NAX just makes sure that conversation only happens with people worth the hour.
Recordings sit in a private bucket with a retention window you set. When the window closes, they are gone — and you can delete a candidate's record on request before that.
Three things. That is the whole job on your side.
You describe what a good answer looks like for each role you hire for. We build the question bank and the rubric around it and show it to you before anything goes live.
One link per candidate, generated one at a time or in bulk from a spreadsheet. You can send them from your own inbox or your existing ATS — nothing has to be replaced.
Open the queue once a day. Watch the interviews the score flagged as worth watching, ignore the rest, and book the ones you like. Most teams spend under twenty minutes on it.
NAX handles the configuration. Your team only reviews the shortlist.
We take your job description and turn it into a question bank and scoring rubric.
Your logo, your language, your consent copy — hosted on a NAX subdomain or yours.
Bulk upload or API. Every candidate gets a single-use link with a reference code.
Scored verdicts land in the HR dashboard as sessions finish. Shortlist and export.
Everything hiring teams ask us before a pilot — setup, accuracy, fairness, privacy and cost.
Between 15 and 30 minutes depending on the role and how many live tasks it includes. Candidates get up to 30 seconds of think time per question, so the pace never feels rushed, and they can start whenever they are ready instead of waiting for a slot.
Yes. Each role gets its own question bank, live tasks and scoring rubric. NAX configures them with you during onboarding, and you can revise a question or reweight a competency at any time without disturbing candidates already in flight.
Scores are anchored to model answers written for your role, not to a generic personality model. In pilots the clear passes and clear fails match the recruiter's own judgement closely; borderline candidates are deliberately marked Review rather than forced into a decision, which is why a human always reads the shortlist.
Always. NAX produces a score, a summary and the evidence behind it — the hiring decision stays with your team. Most clients auto-reject only the clearly unqualified and review everything above that line.
Tab switches, window blur, secondary displays, copy-paste, multiple faces on camera and prolonged silence are logged with timestamps as they happen. Five strikes ends the interview automatically and the full log is attached to the result so a borderline case can be judged in context.
Invite links are single-use and expire on your schedule, the camera runs for the whole session, multiple faces are flagged, and the recording is one continuous take. Optional mobile or WhatsApp verification adds a second identity check before the session starts.
The interview can be conducted in any of 125+ languages and regional accents, including a natural rojak mix of Bahasa Malaysia and English. Your plan determines how many languages are active at once, and the interviewer replies in whichever one the candidate is using.
That is exactly why language is detected per turn and scoring is applied to substance rather than fluency. A candidate can answer in their strongest language and still be compared fairly against everyone else in the same role.
The session pauses, buffers locally and resumes on the exact question. Bitrate adapts to the connection in real time, and the upload is resumable — a dropped line costs a few seconds, not the whole interview.
No. The invite opens in a normal browser on desktop or mobile. A device check tests camera, microphone, speaker, lighting and bandwidth before recording begins, so problems are fixed up front rather than discovered in the transcript.
Recordings, transcripts and logs live in a private, access-controlled bucket with a retention period you set. Consent is captured before the interview starts, access is limited to your reviewers, and records can be purged on request.
Invites can be generated in bulk from a CSV or through an API, and finished results carry a reference code so they can be matched back to your records. Optional add-ons email results to a distribution list or push them to your own endpoint.
Most clients are live within a week: we turn your job description into a question bank, brand the portal, and send the first invites. Pricing is a one-time setup fee plus a monthly plan based on candidate volume and the number of active languages, with optional add-ons billed separately.
Tell us the roles you hire for and we'll configure a pilot for your team.