AI-powered inbox
The Interactions section is the control center where you can monitor and manage all comments received on your social pages. The interface is organized to give you a clear view and complete control over every interaction.

You can reach this section by clicking on the "Comments" menu.
You can filter comments by status, platform and tags.
Additionally, you can also filter the comments received from a specific social profile or a specific agent.
Status of interactions
- All: view all interactions
- Draft: responses generated by AI pending human approval
- Unread: Facebook comments received and not yet read in ReplyZen
- Replied: comments with AI responses already posted
- Hidden: comments hidden by AI
Comments appear in the "Draft" section in these scenarios:
-
Manual approval mode: When you have "Enable Replies with AI" ON but "Post Replies Automatically" OFF, all AI-generated responses require your approval before posting.
-
Confidence-based publishing: When using "High confidence" publishing mode, replies with low confidence scores are saved as drafts for your review, while high-confidence replies are published automatically.
In both cases, you can review, edit, and approve these responses manually from this interface.
Feedback system for AI improvement
ReplyZen includes a feedback system that helps the AI learn and improve from your input. You can provide feedback on moderation actions and AI replies directly from the comment detail view.
How the feedback system works
When viewing a comment in the detail panel, you'll see feedback prompts that appear after moderation actions or AI replies:
For moderation actions:
- "Rate this moderation action to train ReplyZen" (appears when a comment was hidden)
- "Rate this decision to train ReplyZen" (appears when a comment was reviewed but not hidden)
For AI replies:
- "Rate this reply to train ReplyZen" (appears when an AI reply was generated)
Providing feedback
Step 1: Rate the action
- Click the thumbs up button for positive feedback (good decision/reply)
- Click the thumbs down button for negative feedback (poor decision/reply)
Step 2: Add detailed notes (optional but recommended)
- After rating, a text area appears where you can provide specific feedback
- Explain what was good or what could be improved
- Give context about your business or brand preferences
Step 3: Choose training scope
- Agent-specific: Train only the current agent with this feedback
- Global: Use this feedback to improve all agents in ReplyZen
Feedback options and interface
Viewing existing feedback:
- Previously submitted feedback appears as a summary card
- Shows your rating, notes, and training scope
- Includes edit and remove buttons for updates
Editing feedback:
- Click the edit button to modify your previous feedback
- Update your rating, notes, or training scope
- Submit to save changes
Training scope switch:
- ON: "Use this feedback to train agent [name] only"
- OFF: "Use this feedback to globally train ReplyZen"
How feedback improves AI performance
Your feedback is used to train AI models for similar comments and situations:
- Moderation feedback: Helps the AI learn your specific moderation preferences and policies
- Reply feedback: Improves response quality, tone, and relevance for your brand
- Context learning: The AI learns from your business-specific guidance and requirements
- Pattern recognition: Feedback helps identify and handle similar situations more accurately
Best practices for effective feedback
Be specific in your notes:
- Instead of "bad response," explain "too technical for our customer base"
- Rather than "wrong decision," specify "this violates our community guidelines about promotional content"
Provide business context:
- Explain industry-specific considerations
- Clarify brand voice and tone preferences
- Note seasonal campaigns or special policies
Use training scope strategically:
- Choose "agent-specific" for unique brand requirements
- Choose "global" for universal best practices that benefit all users
Managing your feedback
Review and edit:
- All feedback can be edited or removed after submission
- Use the edit button in the feedback summary to make changes
- Remove feedback entirely with the trash button
Feedback persistence:
- Feedback is saved and can be referenced for future improvements
- Original content (comment text, reply text) is preserved with your feedback for context