OnCueOnCue feature guide · Live support
Live AI interview copilot: hear the question, shape your answer
OnCue captures questions from a remote call through your computer's system audio. Its desktop copilot shows a transcript and answer points organized by question. You remain responsible for judging the content and speaking in your own words. Use assistance only where the interview rules allow it.
From a spoken question to an outline
- Choose an audio input in the desktop copilot and grant the required system permissions.
- Read the question as it is transcribed; you can also type or supplement a question manually.
- Review the answer points, then decide how to respond using your own project experience.
The screenshot shows the copilot before a session starts. Transcription and answer speed vary by device, network, and model.
Local data and cloud requests
Resumes, recordings, and interview history are stored on your device. With your own API key, the app connects directly to the selected provider. Hosted mode sends speech and answer requests through the OnCue gateway. Cloud models receive the content needed to complete each request.
Devices and use boundaries
OnCue has macOS and Windows desktop clients. Native system audio capture on macOS requires version 14.2 or later. Enable interview assistance only where the interview or meeting rules permit it.
What to check before starting
First confirm that the meeting rules allow assistance. Select the system audio you want the desktop app to capture, then check that the required permission and input are working. This helps you catch a missing audio source before the interviewer speaks. For native system audio capture on macOS, also check the operating system version requirement.
When a transcript is incomplete
You can type or supplement the question and review the answer points for that question. The outline can remind you of relevant experiences and topics, but it cannot verify facts about your projects or replace your judgment in the conversation. Background noise, network conditions, and the selected model can affect transcription and generation.