Meeting Intelligence: Real-Time Transcription, Notes, and AI Analysis

Meetings are where decisions happen, but they are also where information gets lost. Someone mentions a deadline, another person volunteers for a task, a technical decision is made in passing — and unless someone is meticulously taking notes, half of it disappears by the next day. Meeting intelligence changes this by capturing everything and turning raw conversation into structured, actionable data.

Gonu AI's meeting intelligence system runs as part of the desktop application. It detects your active meeting — whether on Zoom, Google Meet, Microsoft Teams, or any other platform — and starts capturing audio, transcribing speech, and analyzing the conversation in real time.

Automatic Meeting Detection

When you join a meeting, Gonu AI detects the meeting application automatically. It identifies which app is producing the audio — Zoom, Google Meet, Microsoft Teams, Slack Huddles, or Discord — and configures the audio capture pipeline accordingly. You do not need to select anything manually or click a "start recording" button. The system recognizes the meeting context and begins.

The detection engine uses a confidence scoring system. It checks running processes, audio routing, and window titles to identify the meeting platform with high accuracy. If the confidence is low, it asks for a quick confirmation before proceeding.

Real-Time Transcription

Every spoken word in the meeting is transcribed in real time using advanced speech-to-text models. The transcription appears in the Gonu AI overlay as the meeting progresses, giving you a live text feed of the conversation. This is especially useful when audio quality drops, when someone speaks quickly, or when you need to review something that was said moments ago.

The system captures both system audio (what comes through your speakers from the meeting) and your microphone input separately. This dual-stream approach means speaker attribution is more accurate — the AI can distinguish between what you said and what others said.

Speaker Detection and Attribution

Raw transcription without knowing who said what is only marginally useful. Gonu AI's speaker detection identifies individual speakers throughout the meeting and labels each segment of the transcript accordingly. When the meeting ends, you get a transcript that reads like a script — "Alice: We should prioritize the API migration," "Bob: I can take that, it should be done by Friday."

AI-Generated Action Items and Summaries

At the end of each meeting, the AI processes the full transcript and generates a structured summary. This includes an overview of the meeting's main topics, decisions that were made, action items with assigned owners and deadlines (when mentioned), questions that were raised but not resolved, and key technical or business points discussed.

The summary is not a simple extraction of keywords — the AI understands the flow of conversation and produces a coherent narrative of what happened and what needs to happen next.

Stealth Desktop Overlay

The meeting intelligence interface runs in a stealth overlay that is invisible to screen sharing. When you share your screen during a meeting, the Gonu AI overlay does not appear in the shared view. Other participants see only your screen content — they never see the transcription feed, AI suggestions, or meeting notes.

This is achieved through a separate desktop rendering layer that is excluded from the screen capture APIs used by Zoom, Meet, and Teams. The overlay floats on top of everything else on your screen but is invisible to anyone else on the call.

Attend Mode — AI Attends Meetings For You

Gonu AI's most powerful meeting feature is Attend Mode — the AI can autonomously attend meetings on your behalf. When activated, the AI joins the meeting, watches the screen using real-time screenshots, listens to audio via system audio transcription, and can answer questions using your project context. It controls the desktop (mouse, keyboard, window management) to navigate the meeting app and interact as needed.

Every action is narrated in real time so you can monitor what the AI is doing. You can talk to it or type instructions at any time. An instant kill switch (press Esc) stops all automation immediately for safety. Attend Mode uses the same desktop control engine that powers the full automation system — mouse clicks, keyboard typing, window management, and clipboard access — all cross-platform on macOS, Windows, and Linux.

Auto-Join via Playwright

The Meeting Bot programmatically joins Google Meet, Zoom, and Microsoft Teams using Playwright browser automation. You provide a meeting link, and the bot navigates to the meeting, handles the join flow, and starts capturing audio and video. There is no manual setup required — the bot handles login flows, waiting rooms, and permission dialogs automatically.

Real-Time AI Assistance During Meetings

Beyond transcription and notes, the AI can actively help during the meeting. If a technical question comes up that you need context for, you can ask the AI through the overlay and get an answer without leaving the call. If someone references a document or metric, the AI can pull relevant context from your workspace files. If you need to draft a response or summarize a point, the AI generates it in seconds.

The "Ans" button in the overlay triggers instant AI analysis of the current conversation context. It generates a contextually relevant response based on what has been discussed, your uploaded documents, and the meeting topic. Voice output via ElevenLabs TTS lets the AI speak responses aloud with barge-in support.

Post-Meeting Intelligence

After the meeting ends, all data — transcript, summary, action items, and notes — is saved to your session history. You can review meetings later, search across past transcripts for specific topics, and track whether action items have been completed. The session detail page shows everything organized in tabs: Summary, Highlights, Full Transcript, and Notes.

Meetings are synced from the desktop app to the web dashboard via API, so you can access your meeting intelligence from any device. The data stays yours — stored locally and synced only to your own account.

Getting Started

Download Gonu AI, join a meeting, and let the meeting intelligence system do the rest. The free plan includes 2 meetings per day with basic transcription. Upgrade to Pro for unlimited meetings, Attend Mode, full speaker detection, AI summaries, ElevenLabs TTS, and post-meeting reports with follow-up emails.