Page Comparison

Skill is an enabled interaction between Bot and User which can answer a common query or interface with 3rd party systems to fulfill the user request. It contains the details, phrases for how the skill will be triggered, the conversation flow that has the dialogs between the user and virtual agent, and fulfillment of the conversation. The intention of the user’s request is understood by Luma VA by the inherent Natural Language Processing capabilities.

Now, let us understand the basic building blocks of a Skill.

• Attributes are used to gather and hold data, which can be later used to fulfill the user's request.

• Integration in Luma Virtual Agent represents the connectivity mechanisms to external systems and associated web service operations.

• KPI Metrics

• Invocation Type represents the method using which end users can invoke or request for the Skill.

• Conversation Flow is used to build the Conversation flow between the user and your bot. Here we structure the conversation by adding the following conversation items,

• Build and Publish is used to train the bot on conversations and Skills designed in Bot Builder. It ensures that Virtual Agent understands the user intent and processes the services as requested.

You can refer to Creating a FAQ Skill to understand Step by Step instructions to create a Skill.

A GIF image from a Live conversation

You can integrate Luma Virtual Agent with 3rd party applications to fulfill the user requests. Building a skill to log an Incident in ISM

How is Skill Executed in Luma

Now, that we understand the components of a Skill, let us now understand how does Skill execution works.

1. When an End-user initiates a Luma skill, from any of the supported channels using user phrases, NLP Engine processes the phrase and calculates the confidence score of the phrase against existing skills.

2. The score is then evaluated against the thresholds defined for your Tenant, and the best-suited skill is selected for execution.

3. If the skill is invoked through a command, NLP is bypassed, and the skill corresponding to the command is directly identified for execution.

4. Once the skill is identified, prompts defined in the attributes are triggered, to gather information required for the skill. This information captured in Attributes could be passed to Integration/Web Service Operations to fulfill the user's request on external systems.