Meeting Notes ? Action Items ? CRM

What this playbook does

Implement CRM automation for action item tracking, use workflow tools for follow-ups, and leverage AI for meeting note summaries to improve meeting efficiency and follow-through.

Requirements

What you'll need to run this — tools, time, and money.

Required tools

Optional tools

The workflow

The first 3 steps — enough to confirm this is the right approach for your situation.

1. Set up a centralized meetings database Create a single Notion database to store all meeting notes with structured fields for decisions, action items, and project links.
2. Configure AI summarization and action item extraction Use OpenAI API via Make or Zapier to parse meeting transcripts into summaries, decisions, and action items.
3. Automate task creation and CRM updates Set up workflows to automatically create tasks from action items, notify assignees, and update CRM records accordingly.

4. Use OpenAI API to parse meeting content Send the transcript to OpenAI with a prompt to extract structured meeting summaries, decisions, action items with owners and due dates, and open questions.
5. Create a new Notion page with parsed data Map the AI output fields to the Notion database properties and create a new page for each meeting with the structured data.
6. Automate task creation from action items For each action item, automatically create a task in the project management tool or CRM task list, assign it to the responsible person, set due dates, and priority.
7. Send notifications and reminders Configure notifications via email, Slack, or CRM alerts to notify assignees of new tasks and send follow-up reminders for overdue items.
8. Update CRM records with meeting notes Link meeting notes and action items to relevant CRM contacts or deals to maintain a single source of truth for customer interactions.
9. Schedule follow-up meetings automatically Use calendar integrations to schedule follow-up meetings based on action item status or project milestones.
10. Monitor and refine workflow Collect team feedback, monitor task completion rates, and adjust automation rules and AI prompts to improve accuracy and efficiency.

System prompt

You are an AI assistant that extracts structured meeting summaries, decisions, action items with owners and due dates, and open questions from raw meeting transcripts. Format the output as JSON with fields: summary, decisions, action_items (each with description, assigned_to, due_date, priority), and open_questions.

Example

Input:

In today's meeting, we decided to launch the new product by July 15. John will prepare the marketing plan by June 10. We need to review the budget next week. Action items include: John to draft marketing plan, Sarah to update the budget spreadsheet.

Output:

{ "summary": "Decided on product launch date July 15.", "decisions": ["Launch new product by July 15"], "action_items": [{"description": "Prepare marketing plan", "assigned_to": "John", "due_date": "2026-06-10", "priority": "High"}, {"description": "Update budget spreadsheet", "assigned_to": "Sarah", "due_date": "2026-05-20", "priority": "Medium"}], "open_questions": ["Review budget next week"] }

Failure modes

AI may misinterpret ambiguous or poorly formatted transcripts, causing incorrect action item extraction or missing assignments. Integration failures can occur if API limits are exceeded or if CRM fields change. Notifications may fail if user contact info is outdated.

Fallback instructions

Manually review and edit meeting notes and action items in Notion. Verify task assignments and due dates before sending notifications. If automation fails, create tasks and update CRM manually until issues are resolved.

Edge cases

Handling sensitive meetings where auto-recording or AI summarization is disabled. Meetings with multiple languages or heavy jargon may reduce AI accuracy. Action items without clear owners or due dates require manual follow-up.

Related playbooks

Other workflows in the Internal Ops category.