To maximize the potential of your AI agents, consider these best practices when creating and organizing Topics:
Define Clear Objectives:
Establish specific goals for each Topic
Ensure each Topic addresses a distinct area of functionality
Group Related Tasks:
Cluster similar actions and processes within a Topic
Maintain logical coherence in task groupings
Use Descriptive Naming:
Choose clear, intuitive names for Topics
Avoid ambiguous or overly technical terminology
Optimize for Natural Language Understanding:
Include various phrasings for common user queries
Incorporate relevant keywords and synonyms
Provide Comprehensive Instructions:
Detail step-by-step processes for each task
Include handling for potential exceptions or errors
Maintain Consistency:
Use uniform formatting and structure across Topics
Ensure consistent language and tone
Regular Review and Update:
Periodically assess and refine Topic content
Add new relevant tasks and remove obsolete ones
Below are a few examples of Good vs Unclear/Bad topics which will enhance your understanding of how to use the above mentioned best practices:
Example 1:
Unclear/Bad Topic: “Customer Support resolution”
Reason: Too broad and lacks clarity. It does not specify what aspects of customer support it handles, leading to ambiguity for both users and the AI.
Good Topic: The "Customer Support" topic should handle all queries related to troubleshooting common product issues, processing refund requests, providing warranty information, and guiding users on how to contact live agents for unresolved issues. This ensures clarity by grouping related tasks and covering key aspects of customer support.
Reason: Clear objectives and logical grouping of related tasks. It also uses descriptive naming and ensures user-friendly natural language understanding.
Example 2:
Unclear/Bad Topic: Leave Requests approval
Reason: Ambiguous and overly generic. It does not explain whether it addresses policy inquiries, application processes, or status updates, which could confuse users.
Good Topic: The "Leave Requests" topic should guide users on how to check leave balances, apply for different types of leaves (e.g., annual, sick, or parental), track leave approval statuses, and address exceptions like withdrawing or modifying a leave request. This ensures seamless support for employees navigating leave-related processes.
Reason: Descriptive and specific. It clearly outlines all related tasks and potential exceptions, enhancing usability and accuracy.
Example 3:
Unclear/Bad Topic: IT Support and query resolution
Reason: Too vague. It does not clarify which IT issues it handles, leading to inefficiency and potential confusion.
Good Topic: The "IT Support" topic should assist users with password resets, troubleshooting common software or hardware issues, reporting system outages, and submitting requests for IT equipment. It also includes handling escalations for unresolved technical problems.
Reason: Detailed and task-oriented. It uses a hierarchical structure and natural language optimization for common user queries.
Example 4:
Unclear/Bad Topic: Training for employees
Reason: Does not specify what kind of training it covers or how it assists users, leaving room for interpretation and errors.
Good Topic: The "Employee Training" topic should provide information on mandatory training modules, optional skill development programs, enrollment procedures, and tracking completion statuses. It also includes instructions for resolving access issues or seeking assistance with course content.
Reason: Focused and comprehensive. It provides step-by-step instructions and ensures users understand the full scope of the topic.
By following these best practices, you can create well-organized, efficient, and user-friendly Topics that enhance the performance of your UnifyApps AI Agents. Remember, the goal is to create a structured yet flexible framework that allows your AI to handle a wide range of user requests with accuracy and ease.