AI Interviews and Proctoring: CloudHire's Interviewer, Scorecards, and Integrity Checks

What are CloudHire's AI interviews and proctoring?

CloudHire's AI interviews are real-time voice interviews conducted by a configurable AI interviewer, generated per job from your job description, delivered in twelve languages, and recorded server-side with proctoring applied to every recording. Candidates join through a magic link sent by email or WhatsApp with no account required, and recruiters get a scorecard with written reasoning, inline video review, and integrity flags covering identity match, multiple faces, prohibited objects, eye tracking, and duration. Usage is billed in interview minutes, not credits.

This guide covers each piece of that sentence in practical detail: what you can configure, what the candidate experiences, what the scorecard contains, what proctoring actually checks, and what is live versus in beta. It is part of the For Businesses guide hub, and the interviewer itself has a dedicated AI interviewer product page.

A configurable interviewer you can preview

The interviewer candidates meet is yours to shape. CloudHire lets you configure the avatar, voice, name, pronouns, and tone, and preview the result before any candidate sees it. That preview step matters more than it sounds: the interview is many candidates' first live contact with your company, and a two-minute preview catches a mismatched tone before five hundred invites go out. A formal, precise interviewer for a compliance role and a warmer, conversational one for retail sales can run side by side on different jobs.

Every job gets its own interview, generated from the JD

CloudHire does not run a shared question bank. For each job, the interview is generated from the job description plus your basic screening questions, so the conversation probes the skills and context of that specific role. A backend engineering interview asks about the systems named in your JD; a field-sales interview asks about quota, territory, and objection handling. Your screening questions, the non-negotiables like notice period, location, or license requirements, are woven into the same conversation instead of living in a separate form candidates abandon.

For roles where you also want objective skill measurement, generated interviews pair naturally with CloudHire skill tests, with the interview covering depth and communication and the test covering hard verification.

Twelve languages, ten of them Indian

CloudHire AI interviews run in twelve languages: English, Hindi, Telugu, Tamil, Marathi, Bengali, Kannada, Malayalam, Gujarati, Punjabi, Odia, and Spanish. The weighting is deliberate. For businesses hiring across India, forcing every interview into English filters candidates by English fluency rather than job ability, which is exactly wrong for roles where the working language is Hindi, Tamil, or Telugu. Interviewing candidates in their strongest language produces answers that reflect what they know, and widens the effective talent pool for the same role.

Magic-link invites: no candidate account, ever

Candidates are invited by magic link over email or WhatsApp. Clicking the link opens the interview directly, with no account creation, no password, and no app to install. Every invite is logged on the candidate's record, so you can see who was invited, when, and through which channel, and any invite can be resent in one click when a candidate goes quiet.

WhatsApp delivery deserves emphasis for the India market, where WhatsApp open rates far exceed email for non-desk and early-career candidates. The same channels carry the rest of your candidate communication too, covered in the multi-channel candidate outreach guide.

What the candidate experiences, start to finish

From the candidate's side the flow is short: a message arrives on WhatsApp or email, one tap opens the interview page, a quick camera and microphone check runs, and the configured interviewer greets them by name in the language you selected for the job. The conversation typically runs ten to forty minutes depending on how deep you set the interview, and candidates answer by speaking naturally rather than typing or recording one-way clips into silence.

That experience is a hiring asset in itself. Candidates who would never complete a five-page application form or install an assessment app will take a spoken interview from their phone in their own language at 9 p.m. For businesses, that shows up directly as higher completion rates and a wider, more representative funnel, particularly for high-volume, field, and early-career roles where drop-off is usually worst.

Real-time voice, recorded server-side

Interviews are live voice conversations, not asynchronous one-way video recordings. The AI interviewer asks, listens, and follows up in real time, which is why the answers carry signal a form never captures. Every interview is recorded server-side, meaning the recording exists on CloudHire's infrastructure regardless of what happens on the candidate's device: a dropped connection, a closed tab, or a device swap cannot delete or alter the evidence trail. Server-side recording is also what makes proctoring trustworthy, since the analyzed video is the canonical one.

Scorecards with reasoning, not just numbers

Every completed interview produces a scorecard scored against three inputs together: your interview template, the candidate's profile, and the job itself. The scorecard shows written reasoning for its conclusions rather than a bare number, and the interview video is reviewable inline right next to the scores, so a hiring manager who doubts a rating can watch the answer that produced it in seconds. Candidates can hold multiple interviews on their record, useful when someone interviews for two roles or re-interviews after feedback.

Reasoned scorecards change the recruiter's job from transcribing calls to auditing judgments. The scoring approach shares its foundation with the match scoring described on the candidate scoring page, applied to live answers instead of profiles.

Proctoring on every recording

Proctoring is not an add-on tier; every CloudHire interview recording is analyzed. Five checks run on each one:

• Identity match, comparing the on-camera person against the candidate's profile to catch proxy interviewing
• Multiple-face detection, flagging a second person in frame
• Prohibited-object detection, flagging phones, notes, or second screens
• Eye tracking, detecting sustained off-screen reading consistent with scripted answers
• Duration analysis, surfacing anomalous timing patterns

Flags attach to the scorecard with the recording one click away, so a human always makes the final call on ambiguous signals. In beta, live proctoring alerts extend the same checks into real time, raising no-face, multiple-face, and prohibited-object alerts while the interview is still in progress rather than after it ends. Proctoring is one half of CloudHire's integrity story; the other half, verified identity via government ID, is covered in the Cloud-ID verified hiring guide and on the Cloud-ID page.

Timezone-aware scheduling, synced to Google Calendar

When interviews are scheduled rather than taken on demand, CloudHire handles the logistics: scheduling is timezone-aware, so a recruiter in Mumbai booking a candidate in Toronto sees both local times correctly, and confirmed slots sync to Google Calendar automatically. No copy-pasting into calendar events, and no 3 a.m. surprises caused by a missed timezone conversion.

Priced in minutes, not credits

CloudHire measures interview usage in minutes of actual interview time. There is no credit currency to convert and no per-interview token math: a 12-minute screen consumes 12 minutes, a 40-minute technical deep-dive consumes 40. That makes budgeting linear and honest; you can forecast cost per hire as expected interview length times pipeline size, and a batch of short volume screens is priced like what it is, cheap. It also removes the perverse incentive credit systems create, where teams skip interviews late in the month to conserve tokens: minutes scale smoothly with actual hiring activity.

Where interviews fit in the CloudHire pipeline

Interviews sit at the center of the CloudHire workflow: candidates arrive from AI candidate sourcing or your inbound applications, get invited by magic link, interview in their strongest language, and land back in your pipeline as scorecards with proctoring flags attached. Verified candidates carry their interview forward as part of their Cloud-ID, so good interviews compound rather than evaporate.

If you are comparing platforms, the CloudHire vs HireVue guide contrasts real-time voice interviewing with one-way video assessment. To run a pilot interview on one of your live roles, contact the CloudHire sales team; you can preview your configured interviewer the same day.