Starting a live interview session

A live session pairs real-time transcription with on-demand AI answers. The AI listens to your interviewer, suggests answers in your tone, and stays out of the way the rest of the time.

This guide covers the web app flow. The desktop app is identical with one bonus: the AI panel is invisible to screen-share (Stealth mode).

Before you start

• Mic working. You don't have to speak into it; we just need to hear the interviewer through your system audio + your own voice for diarization.
• Browser tab ready. Open your interview's Zoom / Meet / Teams call in another tab. You'll share that tab with InterviewChamp (not with the interviewer — they see only your camera).
• Optional but recommended: Upload your resume to Resume AI first. AI answers tailor better to your background.

Step-by-step

1. Go to Sessions AI in the sidebar
2. Click Start Session in the top right
3. Pick a session type — Interview (default) or Regular Call (less formal, no rubric)
4. Pick a tone (full list — see Tone settings for when to use each):
   • Default — direct, confident. Works for most candidates.
   • Professional — neutral, polished. Banking / consulting / corporate.
   • Concise & Direct — short, no filler. Senior IC / staff roles where verbose = red flag.
   • Highly Technical — depth-first, jargon allowed. Distributed systems / ML / infra.
   • Casual & Friendly — warmer, less rehearsed. Startup / cultural-fit chats.
   • STAR (Behavioral) — forces Situation/Task/Action/Result structure. Behavioral rounds.
5. Auto-Answer toggle — leave OFF for your first session. You'll fire the AI manually with Ctrl+Enter when you want it. Auto-mode fires on every interviewer question, which can be too eager if you want to start answering yourself first. See Auto-Answer.
6. Click Start

Sharing the right thing

Two browser prompts appear in order:

1. Microphone access — allow. The AI needs to hear the interviewer (through your speakers/headphones, which the browser captures via the mic).
2. Screen share — pick the tab that has your Zoom/Meet/Teams call. Share that tab (not the whole screen) and check Share tab audio.

Your interviewer never sees this share — it's a separate browser-API stream feeding only InterviewChamp.

During the session

The screen splits into three panels:

• Left — Transcript. Live captions of who said what. "Interviewer" lines are color-coded.
• Middle — AI Suggested Answer. Streams in when you trigger it. Newest answer appends to the bottom; the panel auto-scrolls so you don't have to chase it.
• Right — Screen Reference. Press Ctrl+Shift+X to capture and analyze whatever the interviewer is sharing (a coding prompt, a JD, a doc).

Firing the AI

Three ways:

Method	When to use
Ctrl+Enter (Mac: ⌘+Enter)	Default — answer the latest interviewer line
Click AI Answer button at the bottom	Same as Ctrl+Enter but mouse-driven
Type a manual message + Ctrl+Enter	Override the question — "Use STAR" / "Be brief" / "Pivot to my Stripe experience"

The answer streams in within ~300ms. Read or paraphrase from your screen.

Ending the session

Hit End in the top right. You'll see your transcript, every AI answer, and a scorecard rubric (Pro+/Lifetime). Everything is saved to Sessions AI for review later.

Tips from the founder

• Don't read aloud. Glance, internalize, then talk in your own voice. The AI gives bullet outlines, not scripts.
• Use Tone for the company's culture. Stripe = Default. McKinsey = Professional. Series-B startup = Casual & Friendly. Behavioral round at any company = STAR.
• Auto-Answer is great after session 2. Start manual to feel the rhythm; flip to auto once you trust it.
• Sub-1-second response only happens on broadband. On flaky wifi, latency can spike to 2-3 seconds; mute your mic for 5 seconds and let the AI catch up.

Common questions

Q: I shared my screen but the AI says "no audio detected." A: When the browser asked which tab to share, you forgot to check Share tab audio. End session, start a fresh one, and look for the checkbox under the tab list in the browser prompt.

Q: I'm sharing my whole screen instead of a tab. Is that OK? A: Works, but less reliable — whole-screen share doesn't capture audio on all browsers. Switch to tab-share (and check Share tab audio) for the cleanest setup.

Q: My interviewer's voice is very quiet in the transcript. A: Mic level + speaker level interact. Crank up your speakers (the AI listens through your speakers via tab-audio). Or wear headphones with a mic — but make sure tab-audio is still captured (some browsers stop capturing if audio routes to Bluetooth).

Q: Can I hide the AI panel mid-session? A: Web — no built-in hide (it's just another tab). Desktop — Ctrl+H toggles the overlay visibility. Ctrl+Shift+H is a global emergency hide that works even when another app is focused.

Q: I'm on a coding screen — can I share the code with the AI? A: Yes — Ctrl+Shift+X captures your screen + the last interviewer question and sends both to the AI. The AI answers grounded in BOTH the question and what's on your screen. See Screen Analysis.

Q: The AI is too verbose / answers are too long. A: Settings → Tone → switch to Concise & Direct. Or open the right-panel Custom Prompt box and type: "Maximum 60 words per answer." Both work mid-session.

Q: My session ended unexpectedly — what happened? A: Most common: hit the 60-min monthly cap (Free), hit the 15-min slab boundary (Free auto-refills next slab if cap not hit), or WebSocket dropped from network blip. Check /sessions → your session card status to tell which.

Q: Where did my transcript go after I clicked End? A: /sessions/[id] — the retrospective view. Linked from the End screen automatically. Also clickable from /sessions (the list page).

Still stuck?

Contact us with category "Something is broken" if a flow above failed. Attach a screenshot — we'll usually reply within a few hours.

Video walkthrough

Coming soon.