Interactive demo

Record, transcribe & analyze meetings

Whisper transcribes the audio, a local LLM analyzes the content and generates an executive summary. All on your server, no cloud APIs.

All data in this demo is mocked. The production version processes real audio with Whisper + Ollama.

Whisper Ollama Flask SQLite MediaRecorder

New recording

Maximum 2 minutes per recording

Architecture

How it works

01
Audio capture

MediaRecorder API records directly from the browser. No apps, no plugins. Audio is sent straight to the server.

02
Local transcription

OpenAI Whisper (medium model) transcribes audio on the server. No cloud APIs, no per-minute costs.

03
LLM analysis

Ollama runs a local 7B model that identifies topics, decisions and action items from the meeting.

04
Executive summary

The same LLM generates a professional summary ready to share via email or official documents.

100% on-premise Your data never leaves the server. No external API dependencies.
~1 minute processing time Whisper medium + Ollama 7B. Scalable with dedicated GPU.
Controlled access Shared activation code. No registration, no friction.
Services

Need to integrate AI into your business?

I design and build text analytics, automatic transcription and LLM agent solutions tailored to your operations. Own infrastructure, no recurring cloud API costs.

30-min call. No pitch, no commitment. Let's talk about what you need.