A key to building effective conversational AI systems like SAP Digital Assistant lies in accurately understanding what users want to achieve. This understanding is captured through the concept of intents — representations of user goals or desired actions expressed through natural language. Properly defining intents is foundational to creating engaging, efficient, and helpful digital assistants.
This article explains the importance of intents in SAP Digital Assistant, outlines how to define them effectively, and offers best practices for aligning them with real user needs in the SAP environment.
In the context of conversational AI, an intent is a classification of the user’s purpose behind an utterance. For example, when a user says, “Show me my leave balance” or “Approve invoice #1234,” the underlying intents are “Check Leave Balance” and “Approve Invoice,” respectively. Identifying intents allows the digital assistant to trigger the correct workflows, fetch relevant data, or respond appropriately.
Start by gathering common user requests and business processes relevant to your SAP environment. Examples include:
Engage stakeholders and end users to understand their priorities and pain points.
Group similar user requests that share the same purpose into discrete intents. For example:
Avoid overly broad or overlapping intents, as they can confuse the assistant.
Use clear, concise, and business-relevant names for intents, accompanied by detailed descriptions. This helps developers and business users understand their purpose and scope.
Example:
Provide diverse examples of how users might express the intent in natural language. This trains the assistant’s natural language understanding (NLU) model to recognize varied phrasing.
Example utterances for Check_Leave_Balance:
Identify key pieces of information (entities) embedded in utterances, such as dates, invoice numbers, or product names, which the assistant needs to extract to fulfill the intent.
Example: In “Approve invoice #1234,” “1234” is an entity of type Invoice_Number.
Link each intent to SAP backend processes via APIs, OData services, or workflows. This connection enables the assistant to perform the requested action or retrieve information in real time.
Regularly test the intent recognition using sample conversations and user feedback. Refine intent definitions and sample utterances to improve accuracy.
| Intent Name | Request Leave |
|---|---|
| Description | Enables users to request time off through SAP SuccessFactors integration. |
| Sample Utterances | “I want to request leave for next week.”, “Apply for vacation from May 10 to May 15.” |
| Entities | Leave_Type (e.g., vacation, sick), Start_Date, End_Date |
| Backend Integration | SAP SuccessFactors Leave Management API |
Defining intents is a fundamental step in building a successful SAP Digital Assistant that understands and fulfills user goals efficiently. By thoroughly analyzing user needs, providing rich examples, and linking intents to backend SAP systems, organizations can create conversational AI solutions that enhance user satisfaction, streamline operations, and drive digital transformation.