Browser audio flows through a streaming transcription pipeline and lands as diarized segments stored against the conversation. Works for meetings, calls, sessions, the room. Zero setup.
A sliding window over the transcript flags terms, entities, questions, and comparisons. It spawns async research jobs and surfaces results without being asked.
Personas — name, icon, system prompt — read every transcript and every research job. Reply in chat across one conversation or all of them. Build personas for the work you actually do.
Cross-conversation search and long retention turn every meeting into compounding context. The thing you reach for when you can't remember whether it was Tuesday's call or last week's session.