Straight answers.

It is the screening, scheduling and interviewing layer. Many teams run it as their whole pipeline; others feed decisions back into an existing ATS.

No. Booking, coding tests and proctoring all run in the browser. Candidates receive links over email or WhatsApp.

Under a few minutes: name the workspace, confirm your interview stages, invite the team and connect Teams or Meet. Every step has a default, so you can skip ahead and adjust later in Settings.

The interface is English. Interviews are transcribed in 30 languages: Arabic, Cantonese, Chinese, Czech, Danish, Dutch, English, Filipino, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Macedonian, Malay, Persian, Polish, Portuguese, Romanian, Russian, Spanish, Swedish, Thai, Turkish, Vietnamese. Code-switching, such as between English and Hindi, is treated as neutral, never as a deduction.

Candidate data is anonymised in code before any model sees it, prompts forbid inference on protected characteristics or emotion, and every score records the model, prompt and rubric that produced it.

Evidence in the résumé against the role’s requirements, on 1–5 rubrics that are anchored to examples. The totals are computed in code, not by the model, and the default bar for a role is 70%.

Yes. Each score ships with its reasoning, the red flags it found, and the model, prompt version and rubric version that produced it, so a decision can be reconstructed later.

A cost-efficient model stack behind a single provider layer; the default is DeepSeek via OpenRouter, with Deepgram for transcription.

In your own Microsoft Teams or Google Meet, created through your workspace’s account.

You offer one to five slots for a panel of up to ten interviewers. The candidate picks one from a link that needs no login; the meeting is created and confirmations go out.

A slot that overlaps any panellist’s existing booking is refused before it can be offered, and again at confirmation. Nobody holds two slots at once.

For Teams, the recording is fetched after the call through Microsoft Graph — no bot joins — then transcribed and scored against the round’s rubric. Transcripts can also be pasted or uploaded.

Python, JavaScript, Java, C, C++, Go, Dart and SQL, judged against hidden test cases in a sandbox.

A bank of roughly 1,550 questions across 11 role families, matched to the job title and balanced by difficulty. You can also write your own with visible and hidden test cases.

Sustained patterns, not moments: a second person, a phone, an identity change, an external voice, repeated tab switches. Blinks and glances are ignored. Evidence is attached to every strike, and nothing rejects a candidate automatically.

The test link survives a closed tab or a changed device; the clock keeps running from the original start. Submitted code is kept even if the judge is briefly unavailable.

Yes — 14 days on every new workspace, with every feature and 25 pipeline candidates. No card is required.

Candidate volume. One pipeline candidate, with every round included — screening, WhatsApp, interviews, transcription, scoring — is the unit. Users are unlimited on every plan, and so is every feature; the plans differ only in monthly candidates and support.

No. Every plan has every feature. A bigger plan buys more candidates and faster support. Enterprise adds what large organisations contract for: a security review, dedicated support and guided onboarding.

You are notified as you approach your plan’s limit, and you can move up a plan at any time. Above 1,000 candidates a month, Enterprise sets a custom volume.

Monthly or annually, in INR, invoiced against the selected plan; annual is two months free. No card is collected in-app. Prices exclude 18% GST. Customers billed outside India see USD pricing.

Annual prepayments are refunded pro-rata for unused whole months. Monthly fees are not refunded once a cycle has started; cancel any time and keep access to the end of it.

Your data is safe. Choose a plan to resume sourcing, scoring and interviewing.

Yes. Every record is scoped to your organisation, and stored secrets such as API keys and tokens are encrypted at rest.

With Google or Microsoft single sign-on, or email and password. If the SSO provider is unavailable, the password path keeps the workspace reachable.

No. Proctoring runs in the browser; only evidence attached to a strike (a frame or a short clip) is kept, and the audio transcript used for voice checks never leaves the browser.

Controls are aligned to SOC 2 principles and formal attestation is on the roadmap — not yet attested, so we don’t claim the certification. The security questionnaire and data-processing terms are available today; see the security page.