Creating a new skill
This section describes creating a new skill in the Luma Bot builder.
Intent-Based Skills (Semantic Skills)
Intent-Based Skills use AI-powered classification to define how a skill is recognized and invoked. These skills are built using structured intent parameters called Classification Details. These components are generated through a guided skill creation workflow, where Luma prompts the user for information and automatically constructs the classification logic. This approach eliminates the need for training data, reduces effort, and enhances skill consistency.
Create Skill
When creating a new Intent-Based Skill, the system initiates an AI-guided process to minimize manual effort and ensure accurate classification.
This is available for the tenants, which are Semantic skill-enabled.
Initiating Skill Creation: Navigate to the Skills section in the Skill Builder and select Create Skill.
User is redirected to a conversation mode where the user is prompted to describe the skill’s function and related use cases. This interaction is critical to generating the classification metadata that defines the skill’s recognition pattern.
Classification Details generation: The skill developer can now manually add the details in the skill details form or use AI to generate the details. The user provides the high-level purpose of the skill. The user is asked to describe the action or outcome the skill is designed to fulfill, using natural language.
For Example: "Help user apply PTO"Luma uses this description to generate a formal Purpose, Trigger Criteria, and Exclusion Criteria from the statement. The information is available in chat and the classification definition in the skill details form. This statement can be manually refined if necessary. Details below are automatically generated:
Skill Name – User-friendly name for the skill. Auto-generated from the provided context; editable by the user.
Skill Identifier – A unique identifier for internal reference.
Classification Details - This information is used to identify the skill. It is divided into three types:
Purpose – defines the core intent or objective the skill is designed to fulfill. It captures the high-level action the skill performs, serving as the foundational reference for how the skill is identified, triggered, and aligned with user needs. This statement is derived from user input and can be refined for clarity, consistency, and functional accuracy.
Trigger Criteria – defines when the skill should be invoked. The criteria is derived from user-provided context and consist of intent, phrases, or keywords a user can use when they want to trigger the skill. This enables natural language-based skill invocation, eliminating the need for manually defined training phrases.
Exclusion Criteria – logical conditions or phrases where the skill should be ignored. These may include situations where this skill should not be triggered or does not apply. These are the negative conditions—user phrases, conditions, or contexts that indicate the skill is not relevant.
Luma guides the user to provide more information to be able to generate as much information as possible.
Users may review and edit the criteria before proceeding. You may use Regenerate using AI to rephrase the classification details.
Now, fill in the other fields on the skill:
Category:
Notes:
Command
Now click on Create to check for any conflicting skills and save the skill.
Conflict Check
Luma proactively detects potential conflicts between newly created skills and existing ones. This AI-based conflict detection is critical for maintaining a clean, unambiguous, and well-structured skill repository. Skills with overlapping or similar purposes may result in ambiguity in skill recognition and incorrect invocations. Such overlaps introduced execution errors and increased support overhead. Additionally, resolving these issues required manual audits, retraining efforts, and ongoing maintenance, making the overall process time-consuming and prone to human error.
With AI-powered conflict detection, Luma performs conflict checks at the time of skill creation or update, ensuring that each new skill is uniquely identifiable and does not interfere with existing ones. Luma checks for overlapping classification details. If a near-duplicate or exact match is identified:
The skill creation is blocked and the developer is notified.
Luma highlights the Reason for the conflict and provides Suggested Actions that the skill developer can act upon.
The Skill Developer may update the classification inputs of the existing skill or the new skill. This ensures that Luma operates with higher precision and clarity, improving the end-user experience and reducing support overhead.
Here is an example:
Phrase-Based Skills
Phrase-Based Skills follow the traditional model, where developers manually define training phrases and tag attributes to train the classification engine. When the tenant is Semnatic enabled, the below Create Skill pop-up screen appears:
On the screen, add the following details:
Specify the Skill Name. The skill name appears in the Welcome Skill suggestions or in a validation message to a user while chatting with Luma.
The system automatically suggests a skill. Identifier, which can be edited at the very time when the skill is being created.
Specify the Category to which this skill belongs. The category is used to filter skills within the Bot Builder interface. You can either select an existing category or create your own category by typing the Category name.
Enter the Image URL that you wish to get displayed. We can even skip this field as it is not mandatory.
Specify the skill Description. Use user-friendly terminology as the description is used as the sub-title when a skill is in the Welcome Skill.
Optionally select Mark as Sub-Skill. Checking this checkbox means that the Skill cannot be triggered directly by a user phrase; instead, it can only be triggered by another Skill.
The status is automatically set to active and can be set to inactive if you need to make the skill inaccessible, i.e., if the skill is not required or is pending changes.
Click Create.
The Skill is now created, and the user is moved to the canvas.
Only one skill model—Intent-Based or Phrase-Based—is supported per tenant. Intent-Based Skills are available for tenants upgraded to version 3.8 or later. A migration option is available for existing Phrase-Based Skills
© 2019 Serviceaide 1-650-206-8988 http://www.serviceaide.com info@serviceaide.com