Back to all days

Day 27 — AI Engineering

AI Meeting Notes & Action Items

The meeting ended 5 minutes ago. Everyone already forgot what was decided.

An MCP server that turns meeting chaos into clarity — searchable notes, tracked action items with owners and deadlines, a decision log you can actually find things in, and follow-up reports that write themselves.

Example Queries

“Who owns the Q2 roadmap decisions from last week?”
“Show me all overdue action items — sorted by priority”
“What did we decide about the client portal architecture?”
“Which team member has the most open action items right now?”
“Give me a follow-up report for this week”
“Search all meetings where security was discussed”

Tools

6 MCP Tools

ToolWhat It Does
search_meetingsSearch past meetings by keyword, type, date range, or participant — returns summaries, attendees, and action item counts.
meeting_summaryFull summary of a specific meeting — attendees, notes, decisions, and action items in one view.
action_itemsList and filter action items across all meetings — by status, owner, priority, or due date. Find overdue items instantly.
decision_logBrowse decisions across all meetings — search by keyword, filter by impact level, and see context and rationale.
owner_workloadSee action item workload by person — open items, overdue counts, completion rates, and who is overloaded.
follow_upsGenerate a follow-up report — overdue items, upcoming deadlines, in-progress work, and recent key decisions.

The System

From Meeting to Accountability

Every meeting generates notes, decisions, and action items. This server connects them so nothing falls through the cracks:

Capture

Meeting notes stored with speaker attribution, timestamps, and topics. Summaries link to the full discussion so context is never lost.

Extract

Action items pulled from each meeting with owner, priority, and due date. Decisions recorded with context and impact level.

Track

Every action item has an owner and a deadline. Filter by status, priority, or person. Overdue items surface automatically.

Follow Up

Generate weekly follow-up reports — overdue items, upcoming deadlines, workload distribution, and key decisions that need attention.

Why This Build

Why AI Meeting Notes?

The average professional spends 31 hours per month in meetings. And the number one complaint isn't that there are too many meetings — it's that nothing happens after them. Decisions get forgotten. Action items live in someone's head. Follow-ups don't happen.

The problem isn't note-taking — it's that notes sit in a doc nobody reads again. What teams actually need is a system that connects what was decided to who owns what and when it's due. One question: “What's overdue from last week?” and you have your standup agenda.

This tool turns meetings into accountability. Search any past meeting, track every action item, and generate follow-ups that actually get things done. Fourth AI Engineering build of the challenge.

Demo Data

What's Inside

15

Meetings

12

Participants

38

Action Items

15

Decisions

22

Meeting Notes

8

Meeting Types

15 realistic meetings across 8 types — sprint planning, retros, architecture reviews, postmortems, 1:1s, standups, cross-functional syncs, and all-hands. 12 team members across Engineering, Product, Design, Marketing, QA, and Sales. 38 action items with owners, priorities, due dates, and status tracking. 15 key decisions with context and impact levels.

Setup

Try It Locally

cd day27-meeting-notes
npm install && npm run seed && npm run build

Then add the config to Claude Desktop and restart.

Want your meetings to actually lead to action?

If your team's decisions keep getting lost between meetings, let's build a system that makes accountability automatic.