How SCRIBES Works: From Meeting Audio to Actionable Insights

Curious about what happens behind the scenes when SCRIBES joins your meeting? Here’s a transparent look at our pipeline — from the moment our bot enters your call to the moment you receive your summary.

Step 1: Bot Joins Your Meeting

When you schedule a meeting with SCRIBES enabled, our bot automatically joins at the scheduled time. It appears as a named participant — “SCRIBES AI” — so everyone in the meeting knows recording is active. We currently support Google Meet and Zoom, with Microsoft Teams support coming soon.

Step 2: Audio Capture

The bot captures the meeting audio in real-time. This audio is streamed directly to our Swiss infrastructure — it never touches any intermediate servers or third-party services.

Step 3: Speech-to-Text Transcription

Our speech-to-text engine processes the audio and generates a time-stamped transcript with speaker identification. We run our own STT models on Swiss infrastructure — we don’t use Google, AWS, or any external transcription API.

Step 4: AI Summary Generation

Once the meeting ends, our AI analyses the full transcript and generates a structured summary: key decisions, action items, discussion topics, and follow-ups. This runs on our own LLM infrastructure in Switzerland.

Step 5: Your Dashboard

Within minutes of your meeting ending, everything is available in your SCRIBES dashboard: the full recording, searchable transcript, and AI summary. You can share, export, or ask our AI assistant follow-up questions about the meeting.

Every step in this pipeline runs on infrastructure we control, in data centres we trust, under laws that protect you. That’s the SCRIBES difference.