Help Center

A typical session takes a few minutes to set up. We recommend doing a quick dry-run on your own device before you send the first link to a real candidate.

1. From the dashboard, click New session and fill in the role title, a short intro the AI will read aloud, your list of interview questions, a scoring rubric, and a duration (15–25 minutes works well for a first-round screen).
2. Save the session. You will land on the session detail page, which shows a unique candidate link at the top.
3. Open that link in a private/incognito window on your own laptop to verify the prejoin screen, mic test, and the AI’s opening question sound right. You do not need to finish the interview—close the tab when you are satisfied, then delete that test session and create a fresh one for the candidate (sessions are single-use).
4. Copy the candidate link into your usual outreach—email, ATS message, or scheduling tool—and send it.
5. When the candidate finishes (or runs out of time), the session detail page refreshes with the full transcript and an AI scoring report. You can revisit it any time.

You do not need to attend the call live. Most teams review reports asynchronously and forward strong candidates to a human round.

Welcome! There is no app to install and no account to create. Plan to be in a quiet room with a stable internet connection a few minutes before you start.

1. Open the link your recruiter sent you on a laptop or desktop with a working microphone. Avoid taking the call on your phone if you can.
2. Enter your name on the prejoin screen and run the mic test—speak for a few seconds and watch the level bar move.
3. When you are ready, click Start call. Your browser will ask for microphone permission—choose Allow. If you accidentally hit Block, see section 4 below.
4. The AI interviewer will greet you, read a short intro, and then ask its first question. Just talk normally; there are no trick interfaces and no camera required.

You can take a beat to think before answering. If something goes wrong technically, refresh the page and click Start call again—your link stays valid until you complete the interview, and the AI will pick up from where the disconnect happened.

The interview runs entirely in your browser. You will need:

• A current version of Chrome, Edge, Safari, or Firefox. We recommend updating to the latest stable release—older browsers often have broken WebRTC audio.
• A working microphone. Built-in laptop mics are fine; a wired headset sounds noticeably better.
• A stable internet connection with roughly 2 Mbps up and down. Anything that supports a normal video call will work. Wired Ethernet beats Wi-Fi; Wi-Fi beats cellular tethering.
• A quiet room. Background noise (TVs, music, open offices) makes it harder for the AI to understand you and harder for the recruiter to review.
• Outbound network access on ports 443 and the WebRTC media ports. If you are on a strict corporate or school network, see section 10.

Mobile browsers can technically work, but we strongly recommend a laptop or desktop. Some mobile browsers reclaim audio focus when notifications, calls, or low-power mode kick in, which can cut the interview short.

If the mic test bar stays empty or the call cannot start, your browser most likely does not have microphone permission. Each browser handles re-granting differently—follow the steps for the one you are using.

Chrome (Windows, macOS, Linux):

1. Click the small lock or tune icon to the left of the URL.
2. Find Microphone in the permissions list and set it to Allow.
3. Reload the page and click Start call again.

Edge:

1. Click the lock icon to the left of the URL.
2. Open Permissions for this site and switch Microphone to Allow.
3. Reload the page.

Safari (macOS):

1. Open Safari → Settings → Websites → Microphone.
2. Find this site in the list and set it to Allow.
3. Also check System Settings → Privacy & Security → Microphone and confirm Safari itself is allowed.
4. Reload the tab.

Safari (iOS / iPadOS):

1. Open Settings → Safari → Microphone and set it to Allow (or Ask).
2. Also check Settings → Safari → Advanced → Website Data if you previously denied this site.
3. Reload the tab.

Firefox:

1. Click the lock icon to the left of the URL.
2. Open Connection secure → More information → Permissions.
3. Find Use the Microphone, clear any block, and reload the page so Firefox can re-prompt.

If permission is granted but the bar still does not move:

1. Check that the correct input device is selected in your OS sound settings.
2. Close other apps that may be holding the mic (Zoom, Google Meet, Discord, Teams, OBS).
3. Unplug and replug your headset, then reload the page.
4. If you are on macOS, toggle System Settings → Privacy & Security → Microphone off and on for your browser.

The AI is good at recovering from background noise, but cleaner audio always produces a smoother interview and a more accurate transcript.

Do:

• Prefer a wired headset with a boom mic over your laptop’s built-in microphone. Wired earbuds with an inline mic are a strong second choice.
• Sit in a quiet, enclosed room. Carpets, curtains, and closed doors all help. Cafés and open-plan offices are tough environments for any voice interview.
• Keep the mic 4–8 inches from your mouth and breathe past it, not into it.
• Mute notifications on your laptop and phone so they do not interrupt the call.

Do NOT:

• Do not use AirPods or other Bluetooth headphones on iOS Safari. This is a known stack issue: iOS frequently downgrades the mic to a low bitrate or drops the input entirely mid-call. Use wired earbuds or your device’s built-in mic instead.
• Avoid Bluetooth headsets in general if you can—they add latency, which can make the AI accidentally talk over you or cut your sentences short.
• Do not run the interview through a virtual mic (OBS, Loopback, Krisp) unless you have tested it—some of them mangle the WebRTC stream.
• Do not put your phone next to your laptop microphone. Touch notifications and haptics show up loudly on the transcript.
• Do not switch audio devices mid-call. If you must, end the call cleanly, swap, and rejoin—your link is valid until you finish.

The AI interviewer is designed to feel like a calm phone screen. A few things to know:

• Take your time. Brief pauses are fine. The AI waits for you to finish before responding, and it will not penalize you for thinking before you speak.
• If you go silent for a while, the AI may gently check in with something like “Are you still there?” or move on to the next question.
• Follow-up questions are normal. If your answer is short or unclear, the AI may ask for an example or a bit more detail—just like a human recruiter would.
• Interruptions go both ways. If the AI starts talking when you have more to say, just keep speaking; it will stop and let you finish.
• If you get cut off mid-call (Wi-Fi blip, browser crash, accidental tab close), reopen the original link and click Start call again within the session window. The interview resumes; the recruiter sees one continuous transcript.
• Time limit. Each interview has a set duration. You will hear a two-minute warning before time is up so you can wrap your final thought.

You can speak in your most comfortable language if the recruiter has configured one—otherwise, English is the default.

When the interview ends:

• If you reach the time limit, the call automatically disconnects after a short closing message. You do not need to do anything.
• You can also end the call early by closing the tab—though we recommend finishing the planned questions if you can.
• The recruiter receives the full transcript and an AI scoring report on their session detail page, usually within a minute or two.
• Candidates do not see their own score. Scores are an internal hiring tool, and the recruiter is the one who decides next steps based on the full report and their own review.

How recruiters should read AI scores. The report breaks each rubric criterion into a score, a short rationale, and the transcript excerpts that drove the score. Treat the number as a starting point, not a verdict—read the rationale and excerpts, and weigh them alongside the candidate’s resume and any human signal. Identical scores can mean very different things in context. If a score feels off, the transcript is the source of truth.

If you have a question about timing or next steps as a candidate, please reach out directly to the recruiter who sent you the link—they own the hiring process.

If you are a candidate, you may understandably want to know how your voice and transcript are handled, and what to do if a voice interview is not the right format for you. The short answers:

• Your audio is processed to generate a transcript and a structured report.
• The transcript and report are shared with the hiring team at the company that invited you—not posted publicly and not sold.
• We do not use your recordings to train general-purpose AI models without explicit permission.
• You can ask the recruiter for a human-led alternative or to delete your data, subject to your local laws.

Accessibility. The interview produces a written transcript of both sides of the conversation, which can be shared with the candidate on request. We do not currently render live captions in-call, so if you are deaf, hard of hearing, or otherwise need captions or a non-voice format, please contact the recruiter before starting and ask for an accommodation—most commonly a text interview, an extended time allowance, or a human-led screen instead. Reasonable accommodations are part of the recruiter’s legal obligation, not a favor.

For the full details, please read our Privacy Policy and our AI Hiring Compliance page.

A few questions we hear often from the recruiter side:

• How do I test a session before sending it? Open the candidate link in an incognito/private window on your own device. Run the mic test and start the call to confirm the intro and first question read correctly. Close the tab when you are satisfied, then delete that test session and create a fresh one to send—each session is single-use.
• Can I edit questions after creating a session? Yes—until the session status changes to in_progress. After that, edits are locked so the candidate’s experience matches the rubric being scored against.
• Can I rerun an interview for the same candidate? No—each session is single-use. If you want to re-screen the candidate (for example, for a different role), create a new session and send the new link.
• How do I delete a session? Open the session detail page and use the Delete session action. This removes the transcript, report, and candidate link.
• Can multiple teammates review the same session? Yes—anyone on your team account can open the session detail page and read the transcript and report.
• What happens if the candidate never opens the link? The session simply stays in scheduled status. You can resend the link or delete the session and move on.
• How do I interpret the AI score? See section 7—treat the per- rubric scores as a starting point, then read the rationales and transcript excerpts before making a decision.

If the call will not start, the prejoin screen hangs, or you get disconnected repeatedly, the most common cause is a network blocking real-time audio.

1. Try a different network. Switch from corporate Wi-Fi to a personal hotspot, or vice versa. If the call works on one and not the other, the original network is blocking WebRTC.
2. Disable VPNs and proxies for the duration of the call. Many corporate VPNs do not route UDP media traffic correctly and will time out the connection.
3. Check your firewall. The interview uses secure WebSockets (WSS over port 443) and standard WebRTC media ports. Strict workplace or school firewalls sometimes block these—ask IT, or take the call from a personal device.
4. Disable browser extensions, especially ad blockers, privacy extensions, and script blockers, then reload. They occasionally block the audio pipeline.
5. Try another supported browser (Chrome, Edge, Safari, or Firefox) if problems persist.

If something is still not working or you cannot find the answer above, contact us:

• Visit our support page for current contact options and response-time expectations.
• Or email us directly at hello@nexprove.com. Candidates: please mention the company name on the interview invite so we can route your question to the right team, and include your browser, OS, and a one-line description of what happened.

For policy questions, see our Privacy Policy, Terms of Service, and AI Hiring Compliance pages.