Skip to case study
Private AI SystemActive

AI Minutes of Meeting System

An end-to-end meeting workflow that records meetings or accepts uploaded audio, generates transcripts with locally hosted AI, produces structured minutes, extracts action items, and assigns responsibilities and due dates to relevant users while keeping AI processing inside the organization's infrastructure.

AI Minutes of Meeting workflow visualization

Why this work mattered

Context

Meeting knowledge can lose momentum between a conversation, its written record, and the people responsible for follow-up. The system was created to connect those stages in one internal workflow.

My responsibility

Designed and built the end-to-end system at KNS Group, covering audio intake, transcription, locally hosted AI processing, structured meeting minutes, and task assignment.

How I approached it

  • Accept automatic meeting recordings or manually uploaded audio.
  • Generate transcripts and process the conversation with locally hosted AI.
  • Transform the discussion into structured minutes and identified action items.
  • Assign responsibilities and due dates when those details are available in the conversation.

Constraints respected

  • Keep AI processing within the organization's own infrastructure.
  • Represent responsibilities and due dates only when supported by the meeting conversation.
  • Connect the output to individual follow-up rather than ending at transcription.

Verified outcome

The resulting workflow moves from conversation to transcription, structured minutes, and individual action items without sending AI processing outside the organization.

Technology and disciplines

Local AISpeech-to-TextOllamaWorkflow AutomationTask Management

Need a system that turns complexity into useful work?

Contact