This article walks you through how to configure the DM AI Chatbot step by step. Please note that this is an add-on feature. Contact your CSM to activate it.
Step 1. Create a new Reply Workflow
Go to Control Panel >> Workflows
Click + New workflow and select Reply
Select AI Chatbot as the workflow type
Step 2. Configure the DM AI Chatbot
Settings
Choose the AI Prompt
Filters
Select the platforms where the bot should reply (for example, Instagram, Facebook, etc.)
Select the assets (profiles, pages, or accounts) the bot should apply to
Only messages coming from the selected assets will trigger the bot.
Define what triggers the workflow:
Keywords: Use
/ONLYfor exact matches.Sentiment: Positive, neutral, negative, or all.
Language: Filter by language.
Tags: Refine with existing tags.
π Tip: Always add exclusions for negative words/emojis to avoid poor matches.
Add additional filters to enhance precision (optional):
Post Types (organic vs ads)
Post ID or URL targeting
Caption keywords
Specific campaign/ad targeting
Use filters to prevent the bot from responding in unwanted situations, for example:
Escalated or sensitive conversations
Messages that should be handled by a human
Specific keywords, users, or message types
This step helps keep replies brand-safe and context-aware.
Step 3. Add knowledge sources
Upload or connect the knowledge the AI should use when responding.
Knowledge can include:
Help center articles
FAQs
Product documentation
Approved URLs or internal resources
Step 4. Add topics and test your bot
Define topics the bot should be able to handle
Test sample messages directly in the workflow to review AI responses
Adjust prompts, rules, or knowledge as needed
Testing helps ensure replies are accurate, on-brand, and useful.
Step 5. Activate the Workflow
Once satisfied, activate β the workflow will now engage new comments/messages matching your setup.
Step 6. Test on the native platform (recommended)
We strongly recommend testing the DM AI Chatbot directly on the native platform (for example, sending a real DM to the selected asset).
Native testing allows you to:
See how replies behave in real conversations
Validate triggers and filters
Catch edge cases that may not appear in preview tests
Examples of Conversations to Automate with DM AI Chatbot
π¬ Collaboration Requests
Keywords:
"collaborate"OR"partnership"OR"collaboration"Sample Response: β"Hey! You're chatting with our bot. Interested in collaborating? Reach out to [email protected]!"
π¬ Shipping Information
Keywords:
"shipping"OR"deliver"/NOT"hate" "dislike" "terrible"Sample Response: β"Here's where we ship! [Link]"
π¬ Order Status Updates
Keywords:
"order status"OR"track my order"/NOT"wrong order" "missing"Sample Response: β"Track your order here: [Link]"
π¬ Refund Requests
Keywords:
"refund"OR"return policy"/NOT"refuse refund" "no returns"Sample Response: β"Learn about our return options here: [Link]"
π¬ Customer Support Requests
Keywords:
"support"OR"help with"/NOT"not helpful" "bad support"Sample Response: β"Here's how to contact our support team: [Link]"









