Scope
- Audience: users who need to quickly try the AI Interview Assistant in the browser.
- Validation goal: confirm that start, transcript, answer, stop, and upload are available.
- Page role: minimum viable flow, without expanding advanced settings details.
Checks before starting
- You entered "Web experience" from the download page and signed in.
- Browser microphone permission is allowed.
- If interviewer audio needs to be recognized, prepare to enable system audio sharing.
First validation configuration
- Auto answer: on, to validate the answer flow directly.
- Knowledge base: off, to exclude retrieval variables first.
- Answer language: Chinese, then compare with English later.
- Speech model: default or team-recommended model.
Standard operation flow
- Enter the "AI Interview Assistant web experience" page.
- Complete the first settings using the configuration above.
- Click "Start" and complete permission and sharing authorization.
- Run a 30 to 60 second simulated Q&A and observe the transcript on the left and answer on the right.
- Click "Stop" and wait for session end and settlement feedback.
Acceptance criteria
- Transcript flow: text keeps landing, not just flashing as a draft.
- Answer flow: at least one complete answer is generated successfully.
- Stop flow: the session ends normally and returns to the ready-to-start state.
- Record flow: when conditions are met, the result can be found on the records page.
Common deviations and handling
- Clicking start returns "Already running": choose to continue the previous session first, or end it and create a new one.
- System audio is enabled but there is no interviewer text: recheck the checkbox in the sharing dialog.
- There is transcript text but no answer: confirm that auto answer is on, or manually click "General answer," "Coding answer," or "Answer this item."