Introduction
Assisting customers with Intelligent virtual agents, often referred to as agents, is a major business trend today. Agents are being used to help ease agent workloads by handling specific types of calls, help with customer information gathering, complaint resolution, executing actions, and helping in other scenarios. Agents help users accomplish specific tasks by using artificial intelligence to identify the customer's intent and present content or execute actions related to that intent. For example, if someone asks what the weather is where they live, an agent could find out where they live and provide them with a detailed weather forecast for their area. That agent could also help the customer do things like check their account balance, book a hotel room, or hand them off to live agent that can better assist them. The agent just needs to know what information to present to the customer based on what they're asking for.
Think of a customer's interaction with a virtual agent just like a conversation you would have with a person. While each conversation is different, they all have their main parts:
Conversation beginning: An event such as answering a phone, a face-to-face greeting, or engaging in other formats initiates each conversation.
Discussion points: These represent the specific items communicated during the conversation such as the weather, making or formulating plans, providing life updates, asking questions and so on.
Conversation ending: The conversation ends with an event such as hanging up the phone, shaking someone's hand, driving away, or some other action.
Conversations are different because what you say or what you do is based on feedback you're getting from the person you're interacting with. For example, if they ask you about your new job, you're going to tell them about your job, not about a vacation you just took. Agents need to be able to work the same way. They need to deliver the correct details and take appropriate action based on what the customer is saying.
Microsoft Copilot Studio’s agents do this through topics. Think of a topic as a small individual conversation on a specific subject. Multiple topics can be used together in a single agent to provide the customer with an automated conversation that feels natural and flows appropriately.
For example, a single agent might contain the following topics:
Greeting
Account Inquiry
Weather Inquiry
Find an Answer
Escalate Conversation
End Conversation
Each of the above topics would have trigger phrases that help the agent identify when to present that topic to a customer. If the customer asks, "What is the weather," the weather topic is launched. If they state, "I have an account question," the account inquiry topic is launched. By defining multiple topics within a single agent, organizations can create flexible virtual agents that can be used to engage and interact with customers on a wide variety of individual topics.
Throughout this module, we examine how to use topics to create and manage conversation paths in an agent.