Browse documentation

Assist Mode

The live AI co-pilot: context, answers, and cue cards.

In Full experience mode, the Assistant panel (the "brain") is help during the meeting, not just a summary after. Real-time answers need a configured provider key (e.g. Gemini); background topic tracking and fast labels can run on small on-device models.

The three tabs

📸 Capture Screen

Grab a screenshot of what's on screen and feed it to the assistant as context.

💡 Context

Background explanations and a Context TL;DR appear automatically as others talk, grounded in your attached context and MCP tools.

💬 Answer

When someone asks a question, suggested answers appear in real time.

Auto-Assist

Toggle Auto-Assist on and the assistant follows the conversation on its own — surfacing context and answers without being prompted. Set it on by default under Settings → General & Recording.

Help Me Answer — cue cards

⚡ Help Me Answer generates Suggested Answers & Cue Cards from the last few turns of the transcript (Alt+A). You can:

  • Expand to sentence — turn terse speaking cues into a full spoken script.
  • Regenerate the cues, and Copy Context or Copy Answer.

🔮 Predicted Questions are speculatively pre-generated and cached, so an answer is ready the moment a likely question is actually asked.

Ask bar

At the bottom, "Ask your assistant…" lets you ask anything about the live meeting; the answer is grounded in the running transcript plus your context. Requires a provider API key.

Was this page helpful? Get RecorderDesk