Automating Meeting Notes with AI and Notion
The article describes a system that uses AI to extract tasks and decisions from meeting transcripts and automatically creates structured Notion pages, helping to avoid important information getting lost in chat history.
Why it matters
This system helps organizations better capture and act on the outcomes of their meetings, improving accountability and productivity.
Key Points
- 1The system sends meeting transcripts to the Claude Haiku AI model to extract key information like meeting title, summary, decisions, and action items
- 2It then creates a structured Notion page with the extracted data, including action items as checkboxes and priority-coded
- 3Notion is used as the core of the system, allowing the meeting pages to be queryable and actionable
Details
The author has been automating meetings for a 236-person company and faced the problem of important decisions and tasks from meetings disappearing into chat history. To solve this, they built a system that takes any meeting transcript, uses AI to extract key information, and automatically saves it as a structured Notion page. The system sends the transcript to the Claude Haiku AI model via the OpenRouter API, which returns a JSON schema with the meeting title, summary, decisions, and action items (including owner, priority, and due date). This data is then used to create a Notion page with the decisions as bullets and action items as checkboxes. By using Notion's database format, the meeting pages become queryable, allowing users to filter open action items across multiple meetings.
No comments yet
Be the first to comment