REAL-TIME GUIDANCE
Follow live interview questions without losing the conversation.
The desktop Companion sends the audio sources you enable to the active session, while the Web workspace shows progressive transcripts and prepares guidance around confirmed interviewer questions.
How does live interview guidance work?
OfferSteady separates microphone audio from supported computer-output audio so the Web workspace can label candidate and interviewer speech. Progressive transcription can change while a sentence is still being recognised; confirmed turns are retained in the conversation history. The assistant uses confirmed interviewer questions together with the context selected for that session.
You can request a Quick Answer manually or enable Auto Answer for confirmed questions. Automatic mode is optional, and neither mode changes the requirement to verify the final suggestion before using it.
What do you need before a session?
Install and open the compatible desktop Companion, grant only the permissions required by the feature, choose the correct microphone and computer-output source, and connect it with the code shown in the preparation page. Confirm that the selected audio meters respond before entering a live interview.
Audio availability depends on the operating system, device route, headset state, conferencing software, and permissions. Changing an output device during a session can require the Companion to recover or reconnect the capture route.
What happens to audio and transcripts?
Raw audio is not stored by default. Audio frames are processed for the active transcription path. Session records and supported transcript data follow the controls described in the Privacy Policy and product interface. Avoid capturing conversations unless you have the required permission or consent.
Why can a live transcript change?
Speech recognition first produces a partial interpretation of the current phrase. That text can grow or be corrected as more audio arrives. A confirmed turn is more stable, but background noise, overlapping speakers, accents, specialised names, and an incorrect audio source can still reduce accuracy.
Use the speaker labels and the original conversation together. If the displayed interviewer question is incomplete, wait for the confirmed turn, enter the question manually, or ask the interviewer to repeat it. Do not treat a partial transcript as a final record.
How do you recover from an audio-route change?
If you connect or remove a headset, switch meeting software, or change the computer's output device, check the Companion's selected sources and meters again. The operating system can move audio to a different route even when the meeting continues normally. Re-select the intended source before assuming that transcription has failed.
If a meter remains inactive, verify the relevant operating-system permission and confirm that another application is not using an unexpected input. Keep the Web session open while restoring the device route, then confirm that new speech appears before requesting guidance.
What should you expect from AI guidance?
Network conditions, speech recognition, model providers, device permissions, and source material can introduce delay or mistakes. Keep listening to the interviewer, ask for clarification when necessary, and do not rely on OfferSteady as the sole means of completing an interview.