Which mode should I use?

When to use it

Use Fact-Checker when someone on the call is going to state numbers, technical details, or claims you have real material to check against — a vendor's spec sheet, a candidate's resume, your own team's architecture doc. It checks what's said against what's actually in your Knowledge Hub, live, as the conversation happens.

• Technical interviews — cross-check a candidate's claims about past projects against their resume or portfolio.
• Vendor and sales calls — catch an uptime, compliance, or capability claim that doesn't match the vendor's own spec sheet.
• Design and architecture reviews — flag an outdated claim about what a service already does before it derails the discussion.

How to set it up

1. Load the material you want claims checked against into the Knowledge Hub before the meeting.
2. Start a new Live Meeting session, and set Assistant Mode to Fact-Checker in setup.
3. Attach that Knowledge Hub pack, then start the meeting — verdicts appear automatically as claims come up, no extra action needed.

Without a Knowledge Hub pack attached, Fact-Checker still runs, but every claim reads "Needs review" since there's nothing to confirm it against — it's at its best with real reference material loaded first.

When to use it

Use Catch-Up Summary for meetings you might not follow start to finish — long workshops, calls you could get pulled away from, sessions where you mainly want the gist, not a full transcript. One keypress gets you a short, topic-tagged recap of everything since you last checked.

• Long discovery or client calls — step out for five minutes, come back, catch up instead of asking someone to repeat themselves.
• Multi-hour planning workshops — check in every 20–30 minutes for a running set of topics, useful later when writing notes.
• Passive attendance — sitting in on a long review where you mainly want to track what was covered.

How to set it up

1. Start a new Live Meeting session, and set Assistant Mode to Catch-Up Summary in setup.
2. During the meeting, press the Catch up button in the overlay (or the shortcut shown next to it) whenever you want a recap.
3. Each recap covers only what's new since your last one — it never repeats itself.

When to use it

The default mode — use it any time you want real-time answers to what's actually being asked or discussed, grounded in your own material wherever you've prepared it. It's the right choice for most day-to-day meetings where you're not specifically fact-checking claims or planning to step away.

• Daily standups — quiet, private answers surfaced as questions come up.
• Client calls — pull up a real answer instead of guessing or scrambling to find a doc mid-call.
• Anything without a more specific need — this is the mode to start with if you're not sure yet.

How to set it up

1. Start a new Live Meeting session — Q&A is the default Assistant Mode, so there's nothing extra to pick.
2. Optionally attach Knowledge Hub material so answers are grounded in your own docs instead of general knowledge.
3. Turn on audio capture and go — questions get answered as they come up.

What it can run

Every interview always includes a shared Technical Q&A round, and you can add any combination of dedicated rounds on top of it — each one gets a fair, real share of the total time you set, not a token few minutes.

• Coding — a real, synced code editor between you and the candidate.
• Low-Level Design — a machine-coding round with a real follow-up conversation grounded in the candidate's own submitted code.
• System Design — a spoken, whiteboard-style design discussion, no editor involved.
• Behavioral — a dedicated round for the softer half of the evaluation.

Setup also lets you set the candidate's years of experience (every question is pitched to that level, not the same for a 2-year and a 12-year candidate), difficulty and depth, and ground the whole thing in a real job description and/or the candidate's resume instead of generic questions.

How to set it up

1. Pick your rounds and set Delivery to Manual, then generate the session and send the candidate the secure Code Editor link for any round that needs one.
2. Watch their code sync live as they write it, with no noticeable delay.
3. Read the AI's private evaluation notes — complexity, correctness, hints you could offer — visible only to you, never the candidate.

How to set it up

1. Pick your rounds and set Delivery to Automate instead — the AI asks each question aloud, listens for the candidate's spoken answer, clarifies without hinting, and paces the whole session against the duration you set.
2. Step back — you don't need to run anything round to round; it moves itself through Q&A into whichever dedicated rounds you selected.
3. At the end, it thanks the candidate for their time and hands you the graded report — exportable as a PDF, with a full question-by-question breakdown, not just a final score.

Want a copy of every finished report kept on your account, viewable from anywhere, not just the device you ran it on? Turn on Settings → Account → Sync session reports — off by default, and covered in our Privacy Policy.

How to use it

1. Open Training Mode and pick a scenario — mock technical interview, sales pitch, or a difficult conversation.
2. Speak or type your responses directly to the AI evaluator in a timed session.
3. Get an instant report card afterward, showing what you covered well and where you fell short.