Understand and edit agent's settings

Getting started with agent settings

When you first create an agent, you'll see four main tabs in its settings panel:

1. Settings: Core interaction and moderation controls
2. Pages: Manage which pages this agent handles
3. Knowledge: Upload and manage information for your agent to reference
4. Testing: Experiment with draft instructions in a safe environment

The Settings tab contains the essential controls for how your agent responds and moderates.

"What settings should I focus on first?"

If you're just getting started, focus on these essential settings:

For interaction (responding to comments):

• Enable Replies with AI: Toggle ON to let your agent generate responses
• Post Replies Automatically: Toggle ON to publish responses without requiring your approval

When both settings are enabled, you can choose between two publishing modes:

• All replies: Publishes every AI-generated response immediately (full automation)
• High confidence: Only auto-publishes replies when the AI is confident, while uncertain replies are saved for your review

Recommended for beginners: Start with only "Enable Replies" turned ON and keep "Post Automatically" OFF. This lets you review and approve all responses before they're published, giving you time to tune your agent's behavior to match your preferences.

Recommended for experienced users: Enable both settings and choose "High confidence" mode. This gives you the best of both worlds—automatic responses for clear situations while maintaining quality control for uncertain cases.

For moderation (handling inappropriate comments):

• Enable Moderation: Toggle ON to activate automatic comment moderation

When this is enabled, your agent will hide inappropriate comments based on your settings.

Common questions about agent settings

"How does confidence-based publishing work?"

When you choose "High confidence" publishing mode, your agent uses AI confidence scoring to decide which replies to publish automatically:

• High confidence replies (above your threshold): Published immediately without your review
• Low confidence replies (below your threshold): Saved as drafts for your approval

You can set your confidence level from four options:

• Permissive (50%): Publishes most replies automatically, only saves very uncertain ones
• Balanced (65%): Good balance between automation and quality control
• Cautious (80%): Only publishes confident replies
• Strict (95%): Only publishes very confident replies

This feature is perfect when you want automation but need quality control for uncertain situations like complex questions, sensitive topics, or ambiguous comments.

"How quickly will my agent respond to comments?"

By default, your agent uses the "Natural" response time, which mimics human behavior with varying delays. You can change this to "As soon as possible" in the advanced interaction settings.

"Will my agent respond in the same language as the comment?"

Yes! By default, your agent automatically detects the language of each comment and responds in the same language.

"Can my agent include links in responses?"

Yes, your agent can intelligently insert links, but it's entirely your choice. You can:

• Allow ReplyZen to automatically select the most relevant link from your post
• Define a specific fixed link to be used
• Choose not to include any links at all

Even if you have multiple different links on the same page, ReplyZen is smart enough to select the most appropriate one for each response context.

Configure these options in the advanced interaction settings.

"How do I control what comments get hidden?"

The default moderation rules hide clearly inappropriate content. For advanced configuration, including custom moderation rules, visit the advanced moderation settings guide.

"Is it mandatory to provide additional context and instructions?"

No, ReplyZen works well with the default settings. Additional context is optional and only needed for specific use cases where you want to give your agent extra information or special instructions.

"Can my agent understand images, videos, and links?"

Yes! ReplyZen includes advanced content understanding capabilities. Your agent can analyze different types of content in posts to provide more relevant and contextually aware replies:

• Image analysis: Understanding product photos, infographics, memes, and other visual content
• Video understanding: Analyzing video content to gain deeper insights into the post context
• Link content: When posts contain links, your agent automatically scrapes and analyzes the linked content (articles, product pages, etc.) to better understand the context for both replies and moderation

This comprehensive content understanding makes automated interactions more meaningful and effective by ensuring responses are relevant to what users are actually seeing and the full context of the post.

Need more control?

If you want to fine-tune how your agent works:

• Advanced interaction settings: Customize tone, response frequency, link handling, and more
• Advanced moderation settings: Configure detailed moderation rules and custom instructions
• Knowledge management: Provide your agent with information to reference in its responses
• Agent testing: Test draft instructions safely before applying them to your live agent