MeetSync is an AI-assisted prototype that converts meeting transcripts into structured summaries, action items, and decisions using LLM-based extraction workflows.
The goal is to reduce post-meeting manual documentation and support async team productivity.
Prototype in progress
(Exploring prompt engineering + extraction workflows)
Input: Raw meeting transcript (Zoom / Meet / Teams / manual paste)
Output: Structured notes with:
- Summary
- Action items (task, owner, due date)
- Decisions
- Discussion points
This helps teams recover key outcomes without reviewing full transcripts.
Phase 1 (Current):
- Prompt-based extraction using LLMs
- JSON-structured output format
- Basic evaluation via notebooks
Upcoming:
- UI dashboard for visualization
- PDF/Markdown export
- (Optional) ASR support via Whisper
Actively using / exploring:
- LLMs (OpenAI API)
- Node.js for backend API (planned)
- Python notebooks for LLM experiments (planned)