Meeting Action Items: Claude + Make

A construction project manager runs three site meetings a week, and every one ends with a dozen things someone agreed to do. They live in his notebook, half-legible, and half of them slip because nobody pulled them into a list anyone checks. The decisions were made; the follow-through is where it falls apart.

This build pulls the to-dos out. Paste the meeting notes or transcript into a sheet, and Claude reads it, extracts each action item with its owner and any due date, and logs them to a tracker. The list that decides whether anything happens writes itself.

We wire it in Make, part of our meeting notes set. This is the simplest shape in the set: a sheet in, Claude in the middle, the same sheet out. The skill is asking Claude for a clean, structured list.

What does this workflow actually do?

In one line: a transcript becomes a task list. Make watches a sheet where you paste notes. Claude reads them and returns the action items with owners and due dates, and Make logs them. The follow-ups buried in a long meeting become a list you can sort and check off.

A few real cases, none of them the usual ones:

• The construction PM above, every site-meeting to-do landing in one tracker.
• A podcast duo pulling next-episode to-dos from their planning calls.
• A graduate student capturing next steps from every advisor meeting.
• A small agency pulling client-call deliverables into a shared list.

The meeting decides what to do; that is the easy part. Capturing each task with an owner so it actually gets done is the part that slips. That extraction is what Claude does cleanly.

Why use Make if Claude can already pull out tasks?

Reasonable question. Claude is good at reading a messy transcript and listing the action items. It is not built to watch a sheet and write the results back into the right place. That is plumbing, and Make does plumbing.

Make handles the no-judgment steps: catching the pasted notes, logging the list, running on its own. Claude handles the one step that needs reading, the extraction. Make is the tracker; Claude is the one who reads the meeting and fills it in. Split that way, the to-dos surface every time without you re-reading anything.

What do you need before you start?

• A Make account. The free 1,000 operations a month covers plenty of meetings.
• A Claude API key from the console, not the chat app. See how to use Claude.
• A Google Sheet with a Transcript column to paste into, plus columns for the extracted items.
• Nothing else. The notes go in a cell; the tasks come out in rows.

Two Make words. A scenario is the whole notes-to-tasks automation. A module is one box. Three boxes, one scenario.

How does the workflow work, step by step?

Three modules, left to right:

Module	App	What it does
1. Trigger	Google Sheets	Fires when you paste a transcript into a row
2. Brain	Claude	Pulls out action items, owners, and due dates
3. Output	Google Sheets	Logs the task list back

Step 1: Set up the meetings sheet

Create a scenario and add Google Sheets, Watch New Rows, connect your account, and point it at a sheet with a Transcript column and a place for the results. Pasting a transcript into a new row starts the run.

Step 2: Let Claude pull the tasks

Add a Claude module, Create a Prompt, paste your key, pick Sonnet:

From this meeting transcript, list every action item.
Transcript: {{Transcript from Step 1}}

Return one line per task, pipe-separated: Task | Owner | Due date
(or "none"). Only include things someone agreed to do. Do not
invent owners or dates that were not stated.

Asking for owner and due date, and forbidding invented ones, is what turns a vague recap into a list you can actually hold people to.

Step 3: Log the task list

Add Google Sheets, Add a Row (or update the same row), and map Claude’s pipe-separated tasks into your columns. Now the meeting’s follow-ups live in a tracker you can sort by owner or due date.

How do you run it and check the result?

Click Run once on a test transcript. When I tested this, the task list came out clean once I told Claude to only include things someone actually agreed to do, which stopped it turning every passing comment into a to-do. Tune the prompt, then turn the scenario on so each pasted transcript yields its task list.

After that, the follow-ups stop living in a notebook. Every meeting produces a sortable list of who-owes-what, which is the difference between a meeting that moves things and one that just happened.

What does this cost to run?

Piece	Free tier	If you outgrow it
Make	1,000 operations/month free	Core plan from about $9/month
Claude API	Pay per use	Pulling tasks on Sonnet costs a fraction of a cent
Google Sheets	Free	Free
Your time	Seconds per meeting	Less as the prompt sharpens

Each meeting is a few Make operations and a sliver of Claude, well inside the free plan. More on tiers in our Make guide.

What can go wrong, and how do you avoid it?

• It invents tasks. Tell Claude to include only things someone agreed to do, the fix that mattered most.
• Owners are wrong or missing. Ask it to leave the owner blank rather than guess; you fill those in.
• The columns misalign. Tell Claude to return only the pipe-separated lines, no preamble.
• One row, many tasks. To split tasks into separate rows, add a Make iterator; start with all tasks in one cell to keep it simple.

The same read-and-extract pattern runs many jobs. See Make AI scenarios.

How do you build this in Zapier or n8n instead?

Same three jobs, different names.

Job	Make	Zapier	n8n
Catch the notes	Watch New Rows	New Spreadsheet Row trigger	Google Sheets Trigger node
Pull the tasks	Claude module	Claude (Anthropic) action	Anthropic node
Log the list	Add a Row	Create Spreadsheet Row	Google Sheets node

Make and Zapier are easiest to start. Zapier vs Make vs n8n compares all three.

Frequently asked questions

Can it send tasks to Asana or Trello?

Yes. Swap or add a final module for your task tool. Starting with a sheet keeps it simple to check before you wire it to a real board.

What if no due date was mentioned?

The prompt tells Claude to write “none” rather than invent one. You add real dates where you need them.

Can I paste rough notes instead of a transcript?

Yes. Claude works from rough notes too; the better the notes, the better the task list.

Can it split each task into its own row?

Yes, with a Make iterator after Claude. The simple version keeps the tasks in one cell, which is fine for most.

Is the chat app the same as the API?

Same models, different door. Make needs an API key from the console, not the chat login.

Sources and references

• Creating a scenario (Make Help)
• Claude API overview (Anthropic)
• Google Sheets basics (Google)
• Claude on Make (integrations)
• Action item — Grokipedia

Last reviewed: May 2026. Make, Claude, and Google update their interfaces; check the official pages for exact button names.

You may also like

• AI Automation for Beginners (the hub)
• Log sales-call notes to your CRM with Claude
• Summarize new documents with Claude + Make
• Make.com Complete Beginner’s Guide
• How to use Claude AI
• Make AI Scenarios: automations that save hours
• The AI Tools Directory