Edge AI
Offline Meeting Transcription
On-device transcription and summarisation for meetings that can never leave the building.
100%
On-premise processing
Zero
Audio leaving the network
The challenge
The client handled sensitive discussions under contractual terms that prohibited sending any audio to third-party cloud transcription services.
Our approach
- 01 Selected and quantised speech models sized to run on local hardware
- 02 Built a diarisation pipeline to attribute speech to individual speakers
- 03 Added summarisation and action-item extraction on the same offline stack
- 04 Packaged the system for deployment inside the client's own network
The impact
Teams gained searchable transcripts and automatic summaries with no audio ever leaving the client's own infrastructure.
