How to Build an AI-powered Q&A Bot from OpenAI to Slack with Zapier

Dashboard > Create Zap > Trigger

Create New Zap with Slack mention trigger

Set up the trigger to fire when your bot is mentioned in any Slack channel. This catches @bot-name mentions and extracts the question text.

1. 1Click 'Create Zap' from your Zapier dashboard
2. 2Search for 'Slack' in the trigger app dropdown
3. 3Select 'New Mention' as your trigger event
4. 4Click 'Continue' to proceed to connection setup

Trigger > Account > Sign in to Slack

Connect your Slack account

Authenticate with Slack to give Zapier permission to read mentions and post replies. You'll need admin permissions to install the Zapier app in your workspace.

1. 1Click 'Sign in to Slack' button
2. 2Select your workspace from the dropdown list
3. 3Click 'Allow' to grant Zapier the necessary permissions
4. 4Wait for the green 'Connected' confirmation message

Trigger > Set up trigger > Configure fields

Configure mention detection settings

Set which channels and mention patterns should trigger the bot. You can restrict to specific channels or allow all channels where the bot is added.

1. 1Leave 'Channel' field blank to monitor all channels
2. 2In 'User' field, select your bot user account
3. 3Set 'Include Bot Messages' to 'No'
4. 4Click 'Continue' to save trigger settings

Trigger > Test > Find new records

Test the Slack trigger

Zapier needs a real mention to configure the data mapping for subsequent steps. Post a test message mentioning your bot in Slack.

1. 1Go to your Slack workspace and post '@bot-name What is API rate limiting?'
2. 2Return to Zapier and click 'Test trigger'
3. 3Wait for Zapier to find your test mention (takes 1-2 minutes)
4. 4Click 'Continue with selected record' when the mention appears

Action > Choose app > OpenAI

Add OpenAI action step

Configure the GPT-4 API call to process the question text from the Slack mention. This sends the user's question to OpenAI and receives the AI response.

1. 1Click the '+' button to add a new action step
2. 2Search for 'OpenAI' in the action app dropdown
3. 3Select 'Send Prompt' as your action event
4. 4Click 'Continue' to proceed to OpenAI connection

Action > Account > Sign in to OpenAI

Connect OpenAI API account

Add your OpenAI API key to authenticate requests. You'll need a paid OpenAI account with API access and available credits.

1. 1Click 'Sign in to OpenAI (GPT-3, DALL-E, Whisper)'
2. 2Paste your OpenAI API key in the 'API Key' field
3. 3Click 'Yes, Continue to OpenAI (GPT-3, DALL-E, Whisper)'
4. 4Wait for the connection test to complete successfully

Action > Set up action > Configure OpenAI

Configure GPT-4 prompt settings

Set up the model parameters and prompt structure. Use GPT-4 for better responses and configure the prompt to include context about being a helpful assistant.

1. 1Set 'Model' dropdown to 'gpt-4' (not gpt-3.5-turbo)
2. 2In 'Messages' field, select 'Custom' option
3. 3Set 'Role' to 'user'
4. 4In 'Content' field, insert the Slack message text using the data picker

Action > Set up action > Response settings

Set response parameters

Configure token limits and response behavior to keep answers concise and cost-effective. Set reasonable limits to prevent overly long responses.

1. 1Set 'Max Tokens' to 300 (keeps responses under ~225 words)
2. 2Set 'Temperature' to 0.7 for balanced creativity
3. 3Leave 'Top P' and 'Stop Sequences' fields empty
4. 4Set 'N' (number of responses) to 1

Action > Test > Send test to OpenAI

Test the OpenAI integration

Send your test question to GPT-4 to verify the API connection and response format. This ensures the AI generates appropriate responses.

1. 1Click 'Test step' in the OpenAI action
2. 2Wait 5-10 seconds for GPT-4 to generate a response
3. 3Review the response text in the test results panel
4. 4Click 'Continue' if the response looks appropriate

Action > Add step > Slack

Add Slack reply action

Configure Zapier to post the AI response as a threaded reply to the original mention. This keeps conversations organized and provides context.

1. 1Click '+' to add another action step
2. 2Search for and select 'Slack' as the action app
3. 3Choose 'Send Thread Reply' as the action event
4. 4Select your existing Slack connection or reconnect if needed

Action > Set up action > Configure message

Map reply message fields

Configure the threaded reply to use the AI response text and reply to the original mention thread. Set up proper channel and thread targeting.

1. 1Map 'Channel' to the channel from the Slack trigger data
2. 2Map 'Thread Timestamp' to the timestamp from the original mention
3. 3In 'Message Text' field, insert the OpenAI response using data picker
4. 4Leave 'Bot Name' and 'Bot Icon' fields empty to use defaults

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

Action > Test > Publish

Test complete workflow and publish

Run the full end-to-end test to verify the bot receives mentions, processes them through GPT-4, and posts appropriate threaded replies.

1. 1Click 'Test step' on the Slack reply action
2. 2Check your Slack channel for the AI response thread
3. 3Verify the response appears as a reply, not a new message
4. 4Click 'Publish Zap' to activate the automation

Drop this into a Zapier Code step.

{{text | replace: '@bot-name', '' | trim}} - What's the best way to approach this?
Context: You're a helpful assistant for a marketing team. Keep responses under 200 words and provide actionable advice.

{{text | replace: '@bot-name', '' | trim}} - What's the best way to approach this?

Context: You're a helpful assistant for a marketing team. Keep responses under 200 words and provide actionable advice.