How to Monitor Slack Feedback for Sentiment with OpenAI and Make

Dashboard > Create scenario > + > Slack

Create New Scenario in Make

Start building your sentiment analysis workflow by creating a new scenario. Make will automatically open the visual editor where you'll connect Slack monitoring to OpenAI analysis.

1. 1Log into Make and click 'Create a new scenario' from your dashboard
2. 2Click the '+' button in the center of the canvas
3. 3Search for 'Slack' and select it from the app list
4. 4Choose 'Watch Public Channel Messages' as your trigger

Slack module > Connection > Create connection

Connect Your Slack Workspace

Authenticate Make with your Slack workspace to access channel messages. Make needs permission to read messages from your #feedback channel specifically.

1. 1Click 'Create a connection' in the Slack module
2. 2Click 'Sign in with Slack' and authorize Make in the popup window
3. 3Select your workspace from the dropdown if you have multiple
4. 4Click 'Allow' to grant Make read access to channels and messages

Slack module > Configuration

Configure Channel Monitoring

Point the trigger at your specific feedback channel. Make will start watching for new messages and pass each one to your sentiment analysis.

1. 1In the 'Channel' dropdown, scroll down and select your #feedback channel
2. 2Set 'Limit' to 10 to process up to 10 messages per scenario run
3. 3Leave 'Types' as default to capture all message types
4. 4Click 'OK' to save the trigger configuration

Canvas > + > OpenAI > Create a Chat Completion

Add OpenAI Module

Connect the OpenAI service to analyze each message that comes through. This module will receive the Slack message text and send it to GPT for sentiment classification.

1. 1Click the '+' button to the right of your Slack module
2. 2Search for 'OpenAI' and select it from the results
3. 3Choose 'Create a Chat Completion' as your OpenAI action
4. 4The modules should now be connected with an arrow between them

OpenAI module > Connection > Create connection

Connect OpenAI Account

Authenticate with OpenAI using your API key. Make needs this to send messages to GPT and receive sentiment classifications back.

1. 1In the OpenAI module, click 'Create a connection'
2. 2Paste your OpenAI API key into the 'API Key' field
3. 3Name the connection something like 'OpenAI Sentiment'
4. 4Click 'Save' to test and store the connection

OpenAI module > Messages array

Configure GPT Sentiment Prompt

Set up the prompt that will analyze each Slack message for sentiment. GPT needs clear instructions to return consistent positive/neutral/negative classifications.

1. 1Set 'Model' to 'gpt-3.5-turbo' for cost efficiency
2. 2In the 'Messages' array, click 'Add item' to create the first message
3. 3Set 'Role' to 'system' and 'Content' to: 'Analyze the sentiment of the following message and respond with only one word: positive, neutral, or negative'
4. 4Click 'Add item' again, set 'Role' to 'user', and map 'Content' to the Slack message text using the data picker

OpenAI module > Advanced settings

Add Response Parsing

Configure OpenAI parameters to get clean, consistent sentiment classifications. These settings ensure GPT returns just the sentiment word without extra explanation.

1. 1Set 'Max Tokens' to 10 to limit response length
2. 2Set 'Temperature' to 0.1 for consistent responses
3. 3Leave 'Top P' and 'Frequency Penalty' at defaults
4. 4Set 'Presence Penalty' to 0

Canvas > + > Slack > Create a Message

Add Slack Response Module

Send the sentiment analysis back to Slack so your team can see the results. This creates a thread reply with the classification.

1. 1Click '+' after the OpenAI module and search for 'Slack'
2. 2Select 'Create a Message' action
3. 3Use the same Slack connection you created earlier
4. 4Set 'Channel' to map to the original channel from the trigger

Slack response module > Text field

Configure Response Message

Format the sentiment result as a clear message that posts back to the channel. Include the classification and maybe an emoji for visual clarity.

1. 1Map 'Text' to combine static text with the OpenAI response: 'Sentiment: [OpenAI response]'
2. 2Set 'Thread TS' to the original message timestamp to create a thread reply
3. 3Leave 'Username' and 'Icon' empty to use Make's default
4. 4Check 'Unfurl links' if you want link previews

📬 New entry: {{1.name}}
Email: {{1.email}}
Details: {{1.description}}

Bottom toolbar > Run once

Test the Complete Workflow

Run a test to make sure messages flow correctly from Slack through OpenAI and back. This verifies your API connections and data mapping.

1. 1Click 'Run once' at the bottom of the screen
2. 2Post a test message in your #feedback channel like 'This feature is amazing!'
3. 3Watch the execution log as each module processes
4. 4Check Slack for the sentiment reply in a thread under your test message

Bottom left toggle > Schedule settings

Enable Real-time Monitoring

Turn on the scenario to start continuous monitoring of your feedback channel. Make will now analyze every new message automatically.

1. 1Click the 'ON' toggle switch in the bottom left of the scenario
2. 2Set the scheduling to 'Immediately' for real-time processing
3. 3Click 'OK' to confirm the activation
4. 4Verify the scenario shows 'Status: Active' in your dashboard

Right-click OpenAI module > Add error handler

Set Up Error Handling

Configure what happens when OpenAI is unavailable or returns unexpected responses. This prevents your scenario from breaking on API errors.

1. 1Right-click the OpenAI module and select 'Add error handler'
2. 2Choose 'Ignore' as the error handling action
3. 3Add a filter to the error handler that catches status code 429 (rate limits)
4. 4Connect the error handler to continue the flow without stopping

Drop this into a Make custom function.

Add this filter between Slack and OpenAI: {{length(1.text) > 10 AND 1.subtype != "bot_message" AND not(contains(1.text; "uploaded a file"))}} This skips short messages, bot posts, and file uploads that waste OpenAI credits.

Add this filter between Slack and OpenAI: {{length(1.text) > 10 AND 1.subtype != "bot_message" AND not(contains(1.text; "uploaded a file"))}} This skips short messages, bot posts, and file uploads that waste OpenAI credits.