Intent Groups

When building advanced conversational bots in Botnex, Intent Groups help you organize and manage related user intents. They make it easier to handle complex conversations where multiple user goals are closely related or should trigger similar bot behavior.

What are Intent Groups?

An Intent Group is a collection of related intents that are grouped together for easier management and logic handling. Instead of treating each intent separately, you can use Intent Groups to trigger the same response or flow for any intent within the group.

This powerful feature helps you create more maintainable bots by reducing duplication and ensuring consistent responses for related user requests.

How Intent Groups Work

Grouping Related Intents: Combine multiple intents that share a common purpose or should lead to the same outcome. For example, "help," "support," and "assistance" can be grouped into a "get_help" Intent Group.

LLM Processing: When Botnex constructs the prompt for intent classification, all intents within a group are included together. The LLM processes the user's message against all individual intents, and if any intent within the group matches, the entire group is considered matched.

Simplified Flow Design: Instead of creating separate branches for each intent, you can connect an Intent Group to a single node or flow, reducing complexity and maintenance overhead.

Flexible User Experience: Users can express their needs in different ways, and your bot will recognize all relevant intents within the group and respond appropriately, creating a more natural conversation experience.

Using Intent Groups in Your Bot

1. Identify Related Intents: Look for intents that should trigger the same response or action in your bot
2. Create an Intent Group: In Botnex, create a new Intent Group and add the relevant intents to it
3. Connect to Flows: Link the Intent Group to the appropriate node or flow. When any intent in the group is detected by the LLM, the bot will follow the same path
4. Maintain and Update: As your bot evolves, add or remove intents from groups to keep your flows organized and efficient

Example: Customer Support Bot

Suppose you have a customer support bot. Here's how you might use Intent Groups:

Intent Group: get_help

• Intents Included:
  • help - "I need help"
  • support - "Can you support me?"
  • assistance - "I need assistance"
  • contact_agent - "Connect me to an agent"
• Linked Flow: Directs the user to support options or connects them with a live agent

Intent Group: order_management

• Intents Included:
  • check_order - "Where is my order?"
  • cancel_order - "Cancel my order"
  • modify_order - "Change my order"
  • track_shipment - "Track my package"
• Linked Flow: Opens the order management interface

Intent Group: account_issues

• Intents Included:
  • reset_password - "Reset my password"
  • login_problem - "Can't log in"
  • account_locked - "My account is locked"
  • update_profile - "Update my information"
• Linked Flow: Starts the account assistance process

This way, no matter how the user asks for help with their account, order, or general support, the bot will recognize their need and respond consistently.

Best Practices

Group by Outcome: Only group intents that should result in the same bot action or response. Don't group intents just because they seem related if they require different handling.

Keep Groups Manageable: Don't overload a group with too many intents. Aim for 3-8 related intents per group for optimal maintainability and LLM processing efficiency.

Review Regularly: As you add new intents, consider if they belong in an existing group or if a new group is needed. Regular reviews help maintain organization.

Use Clear Naming: Name your Intent Groups based on their shared purpose using descriptive, action-oriented names (e.g., "get_help," "order_actions," "account_management").

Test Group Behavior: Ensure that all intents in a group truly should lead to the same outcome by testing various user inputs with the LLM-based classification system.

Document Your Groups: Keep track of why certain intents are grouped together to help team members understand the logic.

Intent Groups Summary

Feature	Description	Example
Intent Group	Collection of related intents	"get_help"
Intents	Individual user goals within the group	"help", "support", "assistance"
Linked Flow	Bot action or flow triggered by the group	Show support options
Benefits	Reduced complexity, easier maintenance	One flow handles multiple intents

Advantages in LLM-Based Systems

Intent Groups work particularly well with Botnex's LLM-based approach because:

• Contextual Understanding: The LLM can understand semantic relationships between grouped intents
• Flexible Matching: Similar expressions are naturally recognized without requiring exhaustive example phrases
• Efficient Processing: Grouping reduces the complexity of prompt construction while maintaining accuracy
• Semantic Coherence: LLMs excel at understanding when different phrasings represent the same underlying intent

By using Intent Groups in Botnex, you can streamline your bot's logic, make your flows easier to manage, and ensure users get the right response no matter how they phrase their requests. Intent Groups are essential for building scalable, user-friendly conversational experiences that feel natural and consistent.