Live coaching

Live coaching is a small card that pops up on the meeting overlay during your call with a short, specific suggestion. It is not a chatbot. It is not a transcript rewrite. One card at a time, either one sentence or a four-section layout. You read it, you act on it or ignore it, and it goes away.

Coaching runs on a loop during the call, roughly every twenty seconds or so, and only when there is something worth saying. It does not fire on every segment. If the conversation is going fine, you will see nothing. The best coaching card is the one that doesn't fire.

• Brief. A single sentence of guidance. This is the default and covers most moments. Example: “Ask what success looks like at 90 days.”
• Structured (four sections). Watch for, Frameworks, Questions to ask, What to avoid. Fires when the conversation hits a topic that genuinely benefits from structure, like a commitment, a negotiation, or a hard objection.

You cannot force a structured card. The engine decides based on what it is hearing. You can decline a structured card if you want to stay heads-down (see Feedback below).

A meeting profile picks which coaching modes are active. Modes include categories like negotiation, objection handling, discovery, hiring, 1:1 management, and others. You pick what you want on for a given meeting by selecting its profile when you start.

To change which modes are on for a profile, open Profiles, pick the profile, and toggle modes on or off. A sales call profile might have discovery and objection handling on. A hiring loop profile might have interview rubrics on and sales modes off.

Coaching suggestions pull from a few sources, in roughly this order:

• Knowledge Base. Your own notes, playbooks, and docs. If you wrote it down, coaching can cite it.
• Meeting Brief. The pre-meeting brief, if you generated one. Context on who is on the call and what the goal is.
• Rehearse pre-meeting prep. If you rehearsed a topic before the call, that prep is available during the call too.
• The recent transcript. The last couple of minutes of what has actually been said.

• Thumbs up. Useful. Keep doing this.
• Thumbs down. Not useful. The engine learns to shut up about that pattern for you.
• Dismiss (x). Not now. No judgment on the pattern itself, just clear the card.

Feedback is private to your account and feeds future suggestions. You can also review every card that fired, with its feedback, on the post-meeting Review page.

• Will not coach you to lie.
• Will not exploit the counterparty's vulnerabilities.
• Will not run covert persuasion patterns.

Full rules are in AI Ethics. If a card ever crosses one of these lines, hit thumbs down and file it in Help → Report. That is a bug.

• Per meeting. There is a toggle on the overlay. Flip it off and no more cards for the rest of this call.
• Per profile. Set defaults in Profiles. A profile can start with coaching off and let you flip it on only when you want it.

• Cmd+Shift+C. Toggle coaching on or off.
• Cmd+Shift+X. Dismiss the current card.

If you are in Self-only privacy mode, coaching still works. It only runs on your words, not the other side's. Suggestions reframe toward your framing, your pacing, and what you may be missing. You will not get suggestions about what the counterparty just said, because the app never heard them.

• Rehearse Mode. Prep for a specific topic before the call.
• Knowledge Base. Add your own notes and playbooks for coaching to cite.
• Meeting Brief. Generate a pre-meeting brief and attach it.
• AI Ethics. The full list of what coaching will not do.