Setup, troubleshooting and feedback

First session

1. Use the desktop package and service address supplied by the operator. Startup checks for updates before sign-in or recovery. Optional updates offer Later; required updates must finish before entry. Reopen the app after installation. If the check fails, retry or contact support.
2. Sign in with Google where account access is required. New accounts ask you to read and accept the published terms/privacy notice. Use the service's access credential only if that setup is offered; provider master keys belong on the operator's server.
3. Choose Online interview for separate microphone and computer audio, or In the room for one microphone. Choose the actual input/output and transcription language. App language and transcription language are separate settings.
4. For a typed first attempt, disable Start listening automatically. Optionally add fictional résumé/job text or a story; select only the documents and examples you want to use. Review an upload preview before Save document; Cancel import discards it.
5. Test audio explicitly captures and sends up to 20 seconds to the speech provider. Test coaching sends a sample question to the model. Neither creates an interview or saves test audio/text in the app; provider charges may apply. Stop test cancels. Changed devices/language/mode invalidate old results.
6. Select Start interview, then send: “How would you make a sample checkout request safe to retry?” Read the accepted answer and check every claim. Use Start listening only when you want audio capture. Missing provider setup, offline samples or an error are not proof of live coaching quality.

Listening, coaching and recovery

• Stop listening stops audio capture. Typed help remains available.
• Pause coaching stops generated guidance while listening/transcription may continue. It does not stop audio capture.
• Hold freezes the selected guidance for reading. It does not pause capture or coaching. Closing the suggestions window also leaves the interview running; Open suggestions window reopens it.
• Pause interview disconnects the live session and retains recovery. Reopen and choose Continue interview to restore the same conversation, with audio stopped until you start listening.
• New interview confirms replacement during a live interview. Startup Start a new interview clears the local recovery pointer after its retention notice. Neither is a guarantee that all server records were deleted.
• End interview & review completes the interview and opens its saved debrief. A completed interview cannot resume. Repeat this question creates a separate practice interview; the original stays completed. Jobs & past interviews lists earlier records.

Solve a problem

• Permission denied: allow microphone access in the OS. macOS online audio also needs Screen & System Audio Recording permission. Reopen when prompted and repeat the audio test.
• Quiet or wrong output: Quiet means no detected speech, not necessarily a fault. Play a sound through the intended meeting output while testing. Select that output, check Bluetooth routing and test microphone/computer audio separately.
• Headset disconnected: stop listening, reconnect/select the device and run Test audio again. Keep the current interview instead of starting a duplicate.
• API unavailable, sign-in expired or rate limited: keep recovery, check the service address, retry/sign in as requested, or wait before retrying a rate limit. Use the operator's final address; service redirects are not followed.
• Speech/model failure: distinguish capture, transcription, API and coaching status. The operator must fix provider credentials/model configuration; never paste provider keys into a support report. Try typed help when only audio is unavailable.
• Wrong speaker or unsupported personal claim: correct the speaker's role/name and retry the answer using corrected context. Review profile facts; an unconfirmed metric must not be represented as verified.
• Insufficient debrief evidence: a typed coaching question or AI answer is not candidate interview speech. A useful report needs attributed final candidate replies. Debrief cues are not a validated performance score or a hiring prediction.
• Export/delete failed: preserve the reference and retry after fixing access. A lost delete response can mean an unknown outcome; verify the record rather than assuming deletion. Canceling an export saves nothing.

Your data

Audio goes directly to the configured speech provider; transcript and selected context go through the API to the coaching provider. Saved PDF/DOCX documents retain original bytes and reviewed text in the API profile; only selected text/facts enter model context. Screenshots are sent to the API/model for that question and Retry answer, with temporary API memory retention; your clipboard, snipping tool or exported image may retain its own copy.

Data and privacy offers Export text, Export printable page and explicit deletion. Exports omit screenshots, setup/profile text, debriefs and practice comparisons; they are not a complete backup. Delete interview removes that server interview, not other interviews, your saved profile, provider copies, backups or files you exported. Delete profile removes profile content separately. Uninstalling the client does not delete server records. Provider/backups retention needs the operator's policy. See the privacy policy.

Compatibility and costs

Windows packaging targets x64 Windows 10 build 19041 or later; diagrams require WebView2. macOS audio targets macOS 13 or later with its native helper. Signed clean installation, Mac capture, exact meeting-app routing and receiver-view privacy acceptance remain pending. There is no certified list of meeting apps/share modes here. Hide from share is a request: verify another participant's view. macOS exclusion is limited; its global shortcuts are unavailable in the current helper.

Published pricing and a trial cutoff are not established. Check account allowances and any provider costs with the operator before starting, including preflight. Usage/token estimates are not an invoice and a request rate limit is not a billing charge.

Feedback and support

For support or feedback, use the existing company contact: [email protected]. Choosing this link opens your email app; this page sends nothing automatically. For privacy requests, see the legal contact information.

Include a short synthetic reproduction with:

• App/package version, OS/build and interface language.
• Online/room/typed mode; device and meeting-app versions if relevant.
• The control used, expected result, actual result and approximate time/time zone.
• Whether capture, transcription, API or coaching failed, and the nonsecret error message.
• Whether retry worked; an optional redacted screenshot with unrelated content removed.

Do not send passwords, access/refresh/session tokens, provider keys, personal résumés or real interview transcripts. No response-time promise is established by this guide.