Step 1: Set up your Notion database

The foundation of our meeting management system is a well-structured Notion database. We'll use it to store meeting information and extracted tasks.

Start by logging into your Notion account and creating a new page titled "Meeting Management".

Inside this page, type "/database" and select the Table view option. This will create a new table database within your page.

Rename the default "Name" column to "Task Name". This column will store the names of our extracted tasks and will be a Text property type.

Next, add the following columns to your database:

1. Assigned to (Text property): To track who's responsible for each task.

2. Task Description (Text property): For detailed information about the task.

3. Timestamp (Text property): To note when in the meeting the task was assigned.

4. Meeting URL (URL property): To link back to the original meeting recording.

5. Status (Select property): To track the progress of each task. Add options like "Not Started", "In Progress", and "Completed".

6. Date (Date property): To mark the date when the transcript/recording is processed.

With these columns set up, our database is ready to receive tasks extracted from your meetings.

Step 2: Set up TLDV for meeting transcription

Now that we have our database set up, let's integrate TLDV to automatically transcribe and summarize our meetings. TLDV is a powerful tool that can join your meetings, record them, and provide AI-generated transcripts.

Head over to tldv.io and sign up for a new account if you haven't already. Click on "Try for free" to get started.

While signing up with Google, make sure to give TLDV access to your Google Calendar events.

During the setup process, TLDV will ask for your meeting recording preferences. You can set it to auto-record all meetings scheduled in your Google Calendar.

Don't worry if you're unsure about the settings now; you can always change them later in the TLDV settings.

Complete the rest of the onboarding steps as prompted by TLDV. Once done, your TLDV account should be set to automatically join and record all meetings scheduled in your Google Calendar.

It's worth noting that you can also manually upload meeting recordings to TLDV. However, keep in mind that there's a limit to the number of free recordings you can upload.

After completing the onboarding process, head to the Integrations section in your TLDV dashboard. Look for the "Manage API keys" option and click on it.

This will take you to a page where you can view and manage your API keys. Locate the Zapier API key - you'll need this for the upcoming Zapier integration.

With TLDV set up, we're now ready to automatically capture and transcribe your meetings. The next step is to connect this with our Notion database and use the Claude API to extract summaries and tasks.

Step 3: Set up Zapier for action item extraction

Our meeting transcripts are being generated by TLDV. Let's now set up a Zapier workflow to extract action items from these transcripts and add them as tasks in our Notion database.

Start by logging into your Zapier account (or create one if you haven't already) then click on "Create" and select “Zap” to start a new workflow.

As soon as you create a Zap, you'll land on a page with "Trigger" and "Action" steps. For the trigger step, search for and select TLDV. Choose "Transcript Added" as the trigger event.

This will automatically trigger our Zap whenever a new meeting is transcribed and added to TLDV. Follow the prompts to connect your TLDV account to Zapier.

Test the Zap to ensure it's working correctly. Zapier should be able to find a recent transcript from your TLDV account.

Next, we'll use the ChatGPT API to extract action items from the meeting transcript. Add a new step and search for ChatGPT. Select "Conversation" as the action event.

To use OpenAI's AI, we need to connect your account. If you don't have an OpenAI account yet, head to https://openai.com and sign up.

Before creating your API key, you'll need to add funds to your OpenAI account. This step is necessary to access the API services. Once you've added funds, navigate to the Dashboard and click on "API Keys" in the left sidebar.

Click "Create new secret key", give it a name like "Meeting Task Extraction”, and copy the generated key. Remember to store this key securely, as you won't be able to view it again.

Back in Zapier, click "Sign in" in the ChatGPT account section and paste your API key. In the ChatGPT setup, select “Conversation” as the action event.

In the ChatGPT step configuration, use the following prompt and map the dynamic value from the previous step, which is the full transcript from TLDV  in the “User Message” field:

Carefully extract all action items from the meeting transcript, including a brief task title in 4-8 words, a detailed task description, the person assigned to each task, the approximate timestamp at which point in the meeting the task was assigned, and a proper reason why this task was given to that person. Format each item as follows:
‍
Task: [brief task title] :: Description: [detailed task description] :: Assigned to: [person] :: Timestamp: [timestamp]
‍
Output only the tasks with all required information, no extra text. And strictly don't include any comma (,) in your response. If any information is unclear or not explicitly stated, note that as "Not specified" in the relevant field. Copy Prompt

For the model, select "GPT-4o mini" as it's well-suited for this task.

Test the step to ensure it's working correctly. Now we have the tasks extracted from the meeting in our desired format. Let's split the data and add each of these tasks as new rows in our Notion database we created.

Add the next action step as "Formatter by Zapier". The action event should be "Text". Use the following configuration as provided in the screenshot below.

Type "Task:" in the separator field - ChatGPT gives us each task using this, so that’s how we’ll identify each separate task.

Now, add a new action step as "Looping by Zapier". With this, we are making sure that each task is split separately to be added to the database recursively.

This allows the zap to run every time within the loop till all the “Task:” values are finished. Set the action event to "Create a loop from text". Use the configuration settings as shown in this screenshot:

In the values to loop, we'll use dynamic data from the previous step of tasks extracted separately by formatter. Make sure you are entering all the information relating to configuration correctly.

Now that we've set up our loop, it's time to extract the specific details from each task. Inside your loop, add a new step and once again select "Formatter by Zapier" with "Text" as the action event. In the configuration tab, we're going to use a feature called "Extract Pattern" to pull out the exact information we need.

In the "Transform" dropdown, choose "Extract Pattern". For the "Input" field, use the dynamic data from Step 4 by clicking on the "Insert Data" button and selecting the output from Step 4. Next, you'll see a "Multiline" option - make sure to set this to "Yes". This ensures that our pattern can match across multiple lines, which is crucial for capturing all the information in longer task descriptions.

Now, here's where the magic happens. In the "Pattern" field, paste the following:

(.*?) :: Description: (.*?) :: Assigned to: (.*?) :: Timestamp: (.*) Copy Prompt

Let’s now instruct the Zap to add all of these separated task details to a new Notion database row in our database.

Add a new step, with Notion as the app and “Create Database Item” as the action. Sign in with your Notion account and give Zapier access to the page of the database we created earlier.

Now, in the configure tab, map the different fields from the relevant results of the previous steps. For the Meeting URL field, map the recording link from the first step of the Zap. Continue mapping out the other fields as shown in this screenshot:

For the Date field, set the start date as "Today's Date" and leave the end date empty. By adding this, Zapier automatically adds the current date when a task is created. Set the “Include time with date fields” as False to keep things clean.

After you are done, test the step, and if you have followed the tutorial up to this point, you have your customized AI meeting management built right inside Notion.

Step 4: Organize your Notion database for efficient task tracking

Now that we have our automated system set up, let's organize our Notion database to make it easy to track tasks and meeting summaries.

In your Notion database, click the "+" button next to the default view name (likely "Table") and select "Calendar" from the drop-down list. This will create a new Calendar view of your database.

This will display all your tasks on the calendar according to their assigned dates, giving you a clear visual overview of when each task was created and allowing you to easily track deadlines and manage your workload over time.

Next, create a Board view by clicking the "+" button again and selecting "Board".

For the Board view, you'll get a visual snapshot of your task pipeline. This layout helps you track progress and manage workflows efficiently.

With these views set up, you now have a comprehensive system for managing your meetings and tasks:

1. The Table view shows all tasks extracted from your meetings.

2. The Calendar view displays tasks based on when they were assigned.

3. The Board view allows you to track the progress of tasks.

This setup allows you to quickly review meeting outcomes, track action items, and monitor progress over time, all within your Notion workspace.

You've successfully built an AI-powered meeting management system using TLDV and Notion. This system automatically transcribes your meetings, extracts action items, and organizes everything in an easily accessible format.

‍

‍

‍

This tutorial was created by Tanmay.