Silent Agent vs Voice Agent: When to Use Each Mode

A brainstorm with five people has different needs than a one-on-one with a client. That's why Vernix gives you two modes: Voice and Silent.

Both join your call, transcribe everything, build a searchable memory, and generate summaries. The difference is how they interact during the meeting.

Voice mode: an active participant

In voice mode, Vernix joins with a microphone. Say "Vernix" during the call, and it responds out loud — pulling context from the transcript, your knowledge base, and connected tools.

Best for:

• Internal team meetings where everyone knows the agent is there. Ask it to recall what was decided in last week's standup, or pull up a spec from your knowledge base.
• Working sessions where you need data fast. "Vernix, what did the client say about the timeline in our kickoff call?" — answered in seconds, without leaving the conversation.
• Onboarding calls where a new team member can ask the agent for context instead of interrupting the flow.

Voice mode uses on-demand activation. Vernix only listens for its wake word. When you say it, the agent connects to respond, then disconnects after 15 seconds of silence. It's not always-on — it activates when you need it.

Silent mode: invisible, chat-based

In silent mode, Vernix joins without a microphone. No audio output, no visible participation. It watches the transcript and responds in the meeting's text chat when someone mentions "Vernix."

Best for:

• Client-facing calls where you don't want to explain why there's a bot on the line. The agent sits quietly, and your team can ping it in chat without the client noticing.
• Large meetings like all-hands or town halls. Nobody wants a bot interrupting. But if someone in chat asks "Vernix, when was the last time we discussed this?" — they get an answer.
• Sensitive discussions where recording audio output would be inappropriate, but having a transcript and searchable memory is still valuable.

Silent mode still transcribes, still builds your meeting memory, and still generates a summary at the end. It just keeps its responses in the chat sidebar.

Quick comparison

| | Voice | Silent | | ----------------------- | -------------------- | --------------------- | | Transcription | Yes | Yes | | Summary generation | Yes | Yes | | Searchable memory | Yes | Yes | | Knowledge base access | Yes | Yes | | How it responds | Out loud | In meeting chat | | Activation | Wake word ("Vernix") | Mention in transcript | | Visible to participants | Yes (as a bot) | Minimal (chat only) |

How to choose

Ask yourself one question: Would everyone on this call be comfortable with an AI responding out loud?

If yes, use voice mode. You get faster, more natural interactions. If not — or if you're unsure — use silent mode. You still get the full transcript, memory, and summary. You just interact through chat instead of voice.

You set the mode when creating each meeting, so you can switch between calls. No global setting to worry about.

Both modes, same memory

Regardless of which mode you pick, every call feeds into the same searchable archive. After the meeting, you can search across all your past calls and pull up context — whether the original call used voice or silent mode.

The mode changes how the agent behaves during the call. What you get after is the same.