How to Clean Up Meeting Notes with Power Automate

My flows > + New flow > Automated cloud flow

Create automated cloud flow

Go to make.powerautomate.com and sign in with your Microsoft account. Click 'My flows' in the left sidebar, then '+ New flow' at the top. Select 'Automated cloud flow' from the dropdown. Name your flow 'Meeting Notes Cleanup' and search for Slack in the trigger list.

1. 1Click 'My flows' in the left navigation
2. 2Click the '+ New flow' button
3. 3Select 'Automated cloud flow'
4. 4Enter 'Meeting Notes Cleanup' as the flow name
5. 5Search for 'Slack' in the trigger search box

Trigger > Slack > When a message is posted to a channel

Configure Slack message trigger

Select 'When a message is posted to a channel' trigger from Slack. You'll need to sign in to your Slack workspace. Choose the specific channel where meeting notes get posted. Set up a condition to only trigger when messages contain specific keywords like 'Meeting:' or 'Attendees:' to avoid processing every message.

1. 1Click 'When a message is posted to a channel'
2. 2Click 'Sign in' and authorize Power Automate to access Slack
3. 3Select your workspace from the dropdown
4. 4Choose the channel where notes are posted
5. 5Click 'Show advanced options' and add keyword filters

Actions > Control > Condition

Add message content condition

Click '+ New step' and search for 'Condition' in the actions. Add a condition that checks if the message text contains your meeting note keywords. Use the 'contains' function with the message text from the Slack trigger. This prevents the flow from running on regular chat messages.

1. 1Click '+ New step' below the Slack trigger
2. 2Search for and select 'Condition'
3. 3Click the left value box and select 'Text' from Slack dynamic content
4. 4Set the condition to 'contains'
5. 5Enter 'Meeting:' or your keyword in the right value box

Condition Yes > + Add an action > HTTP

Add OpenAI connection

In the 'Yes' branch of your condition, click '+ Add an action'. Search for 'HTTP' and select the HTTP action since OpenAI doesn't have a native Power Automate connector. You'll make a POST request to the OpenAI API endpoint. Set the method to POST and the URI to 'https://api.openai.com/v1/chat/completions'.

1. 1Click '+ Add an action' in the Yes branch
2. 2Search for 'HTTP' and select the HTTP action
3. 3Set Method to 'POST'
4. 4Enter 'https://api.openai.com/v1/chat/completions' as URI
5. 5Click 'Show advanced options'

HTTP Action > Headers

Configure OpenAI API headers

In the HTTP action, add the required headers for OpenAI authentication. Add 'Content-Type' with value 'application/json' and 'Authorization' with value 'Bearer YOUR_API_KEY'. Replace YOUR_API_KEY with your actual OpenAI API key from platform.openai.com. Store the API key securely in a variable or use Azure Key Vault.

1. 1Click 'Add new parameter' and select 'Headers'
2. 2Add header 'Content-Type' with value 'application/json'
3. 3Add header 'Authorization' with value 'Bearer sk-your-api-key'
4. 4Replace 'sk-your-api-key' with your actual OpenAI key
5. 5Consider using a variable for the API key

HTTP Action > Body

Build OpenAI request body

In the HTTP action Body field, create a JSON payload for OpenAI. Include the model (gpt-3.5-turbo), messages array with your formatting prompt, and the original Slack message text. The prompt should instruct GPT to format messy notes into structured sections with attendees, topics, decisions, and action items.

1. 1Click in the Body text area
2. 2Paste the JSON structure for OpenAI chat completion
3. 3Set model to 'gpt-3.5-turbo'
4. 4Insert the Slack message text using dynamic content
5. 5Include clear formatting instructions in the system message

Actions > Data Operations > Parse JSON

Parse OpenAI response

Add a 'Parse JSON' action after the HTTP request to extract the formatted text from OpenAI's response. Click '+ New step', search for 'Parse JSON', and select it. In the Content field, select 'body' from the HTTP action. Generate the schema by using a sample OpenAI response or manually define it.

1. 1Click '+ New step' after the HTTP action
2. 2Search for and select 'Parse JSON'
3. 3Click Content field and select 'body' from HTTP dynamic content
4. 4Click 'Generate from sample' or enter the schema manually
5. 5Include choices[0].message.content in your schema

Actions > Slack > Post message

Post formatted notes to Slack

Add another Slack action to post the cleaned-up notes back to the channel. Search for 'Post message' in Slack actions. Select the same channel as your trigger. In the message field, select the parsed content from OpenAI. Add a prefix like 'Formatted Notes:' to distinguish it from the original message.

1. 1Click '+ New step' after Parse JSON
2. 2Search for 'Slack' and select 'Post message'
3. 3Choose the same channel as your trigger
4. 4Click in Message field and select parsed content from OpenAI
5. 5Add a prefix like '📝 Formatted Notes:' for clarity

HTTP Action > ... > Settings > Configure run after

Add error handling

Click on the HTTP action and select 'Settings' from the three-dot menu. Enable 'Configure run after' and check 'is failed' and 'has timed out'. Add a Slack message action in the failure branch to notify users when OpenAI is unavailable or returns an error.

1. 1Click the three dots on the HTTP action
2. 2Select 'Settings'
3. 3Click 'Configure run after'
4. 4Check 'is failed' and 'has timed out'
5. 5Add a Slack message action for error notifications

Flow > Save

Test and save the flow

Click 'Save' in the top right to save your flow. Go to your Slack channel and post a test message with meeting notes that match your trigger keywords. Check the flow run history to see if each step executed successfully. Verify the formatted output appears in Slack within 30-60 seconds.

1. 1Click 'Save' in the top toolbar
2. 2Navigate to your configured Slack channel
3. 3Post a test message starting with your keyword
4. 4Return to Power Automate and check 'Run history'
5. 5Verify the formatted message appears in Slack

Add this JSON body structure to your HTTP action for better OpenAI formatting. Paste this in the Body field of your HTTP request to OpenAI.

{
  "model": "gpt-3.5-turbo",
  "messages": [

{
  "model": "gpt-3.5-turbo",
  "messages": [
    {
      "role": "system",
      "content": "Format messy meeting notes into structured format with sections for Meeting Topic, Date, Attendees, Key Topics, Decisions, and Action Items. Use markdown formatting and bullet points. Be concise but don't lose important details."
    },
    {
      "role": "user",
      "content": "@{triggerBody()?['text']}"
    }
  ],
  "max_tokens": 1500,
  "temperature": 0.3
}