Smart Meeting Summariser & Action-Item Extractor
Turning Teams transcripts into bilingual minutes and tracked action items — so decisions stop falling through the cracks.
Before
Meeting minutes were inconsistent and action items routinely got lost across departments — follow-through sat around 60%.
After
A pipeline summarises Teams transcripts in Arabic and English, extracts owners and deadlines, and posts structured items to Planner. Adopted across 6 departments and ~120 meetings/month, it cut minute-writing time 85% and lifted action-item follow-through to 92%.
Meeting minutes were inconsistent and action items routinely fell through the cracks across departments.
Pipeline ingests Teams transcripts, summarises in Arabic + English, extracts owners and deadlines, and posts structured items to Planner.
- 1
Transcript ingestion
Phase 1Wired Whisper + Teams transcripts into a clean pipeline, handling the reality of mixed Arabic/English speech in the same meeting.
- 2
Structured extraction
Phase 2Moved from a prose summary to structured owner + deadline extraction — the difference between 'nice notes' and something that actually drives follow-through.
- 3
Close the loop into Planner
Phase 3Posted extracted action items straight into Planner via Graph/Power Automate, so the output lands where the work already happens.
Extract structured action items, not just summaries
Why · A summary is read once and forgotten. Owner + deadline items posted to Planner are what moved follow-through from ~60% to ~92%.
Bilingual by default
Why · Meetings mix Arabic and English; producing both keeps every stakeholder able to act without a translation step.
- Adopted across 6 departments and ~120 meetings/month
- 85% reduction in time spent writing minutes
- Action-item follow-through up from ~60% to ~92%
- 01 · The value isn't the summary — it's the tracked action item in the tool people already use.
- 02 · Push outputs to where work happens (Planner) instead of a document nobody reopens.
- 03 · Handle mixed-language audio explicitly; don't assume clean single-language input.
Ask anything about Smart Meeting Summariser & Action-Item Extractor
AI scoped to this project · GPT-OSS 120B