Skip to main content

AI Auto-Responses

Automatic intelligent response system for frequently asked questions using generative AI.

Required Tier

ULTRA - This feature requires Server Premium ULTRA

Description​

The bot automatically detects when a user asks a question in configured channels and generates a personalized response based on the server information you've configured.

AI Model: OpenAI GPT-4o-mini (latest generation)

Features​

Intelligent Question Detection​

  • AI analyzes every message to determine if it's a question
  • Detects questions with or without question marks
  • Examples: "how do I verify", "where is the help channel", "what commands are there"

Configurable Knowledge Base​

  • Write all important information about your server
  • The AI uses this information to generate responses
  • Update the knowledge base when things change in your server

Contextual Responses​

  • AI generates unique responses for each question
  • Friendly and casual tone
  • Mentions specific channels, roles, and commands
  • Short responses (3-4 lines)

Analytics and Feedback​

  • Reaction system (thumbs up/down) to measure effectiveness
  • Statistics for total, helpful, and unhelpful responses
  • Record of unanswered questions to improve the knowledge base

Advanced Configuration​

  • Select specific channels where the system is active
  • Ignored channels where it won't respond
  • Configurable cooldown per user (prevents spam)
  • Option to mention or not mention the user in the response

How to Configure​

1. Access the Panel​

Go to panel.yumechanbot.com → Your server → AI Auto-Responses

2. Enable the System​

  • Toggle "System Enabled" to ON
  • Make sure you have ULTRA tier

3. Configure Knowledge Base​

Write all important information about your server in the editor. Example:

GAMING SERVER - Gamer Community

VERIFICATION:
- Go to #verification
- Click the button
- Accept the rules in #rules

ROLES:
- Game roles in #roles (react to get them)
- Color roles in #colors
- Level 10+ = Veteran Role

ECONOMY:
- /daily - Daily coins
- /work - Work every 4h
- /shop - Server shop

MUSIC:
- /play [song] - Play
- /skip - Skip
- /queue - View queue

IMPORTANT CHANNELS:
- #announcements - Official news
- #help - Staff support
- #general - Main chat

RULES:
1. Respect between members
2. No spam
3. No NSFW content
4. Use appropriate channels

4. Select Channels​

  • Choose channels where the bot will respond (e.g.: #help, #support, #faq)
  • If you don't select any, it will work in all channels

5. Adjust Settings​

  • Cooldown: Time between responses to the same user (recommended: 60s)
  • Mention User: ON to @mention, OFF to respond without mention
  • Feedback: Enable thumbs up/down reactions to measure effectiveness

Best Practices​

Do​

  • Use emojis to organize information
  • Be specific with channel names and commands
  • Include frequently asked information
  • Update regularly when things change
  • Use #channel format to mention channels

Avoid​

  • Sensitive or private information
  • Unverified external links
  • Text that is too long (max 2000 characters recommended)
  • Outdated information
  • Confusing or contradictory instructions

Analytics​

View Statistics​

In the panel you'll see:

  • Total Responses: How many times the AI has responded
  • Helpful Responses (thumbs up): Positive feedback
  • Unhelpful Responses (thumbs down): Negative feedback
  • Effectiveness: Percentage of helpful responses

Unanswered Questions​

The system records questions it couldn't answer (because they weren't in the knowledge base). Use them to improve your knowledge base.

Troubleshooting​

Bot doesn't respond to questions​

  1. Verify it's enabled in the panel
  2. Check the channel is in the allowed channels list
  3. Review the knowledge base has relevant information
  4. Check the cooldown for the user isn't active

Responses aren't accurate​

  1. Update the knowledge base with more specific information
  2. Add context about commands, channels, and roles
  3. Be clear in your knowledge base instructions

Too many incorrect responses​

  1. Increase the cooldown to avoid excessive responses
  2. Review unanswered questions in analytics
  3. Improve the knowledge base with that information

Usage Examples​

Example 1: Verification Question​

User: "how do I verify?" Bot: "@User, to verify go to #verification and click the button. Then accept the rules in #rules and you're done!"

Example 2: Commands Question​

User: "what economy commands are there" Bot: "@User, we have several economy commands: /daily for daily coins, /work to work every 4 hours, and /shop to see the server shop."

Example 3: Channels Question​

User: "where do I see announcements?" Bot: "@User, official announcements are posted in #announcements. You can also find staff help in #help."

Costs​

This feature uses GPT-4o-mini with an approximate cost of $0.00015 per generated response (~0.015 cents). The system is optimized to be economical and efficient.

Privacy​

  • The AI only has access to the knowledge base you configure
  • It does NOT have access to historical server messages
  • It does NOT store personal user information
  • It only processes the current question to generate a response

Support​

If you have issues with AI Auto-Responses:

  1. Review this documentation
  2. Verify your premium tier (requires ULTRA)
  3. Contact support on the official Yume-chan server