Microsoft Copilot Studio: Revolutionizing the Tech World

What is Microsoft Copilot Studio??

Microsoft Copilot Studio is a service that allows you to build AI-powered copilots spanning everything from simple frequently asked questions to more complex troubleshooting that involves lengthy and multi-turn conversations. You can use Microsoft Copilot Studio to build, manage, and deploy copilots that can chat with customers and employees in natural language across many different channels.

Why use Copilot Studio?

Microsoft Copilot Studio offers several advantages for creating interactive and intelligent copilots, including:

1. Easy and Intuitive Authoring: A Copilot author typically develops multiple individual topics, each corresponding to a specific issue or action. Creating topics can be time-consuming and often necessitates specialized knowledge in conversational user experiences, as well as deep technical or domain expertise. These resources are not always readily available in every organization. Copilot Studio makes these resources readily available for all using the "Create with Copilot" feature within Copilot Studio. With this feature you can simply describe your desired outcome, and then AI generates a topic path to achieve that goal. This is made possible by integrating natural language understanding models with Azure OpenAI all in one easy to use platform.
2. Powerful AI Capabilities: Utilize the advanced natural language processing and generation abilities of Microsoft Cognitive Services to enable your copilot to understand and respond to user requests in various languages. Additionally, you can enhance your copilot's functionality and intelligence by integrating it with other Azure services like Speech, Vision, Search, and Knowledge.
3. Flexible and Scalable Deployment: Deploy your copilot to any channel supported by the Azure Bot Framework, including websites, mobile apps, Facebook, Microsoft Teams, and more, offering a versatile and scalable solution.

Lets look at creating a Copilot

One use case for a custom copilot is to create an interface for your customers to ask common questions. For example, you could create a topic within the copilot about your store’s opening hours, named “Store Hours.” When a customer asks a question such as “When do you open?” or “What are your opening hours?”, the copilot uses Natural Language Understanding (NLU) to determine the intent behind the question and matches it to the most relevant topic, in this case, the “Store Hours” topic.

The copilot then follows a conversation flow, consisting of a series of connected nodes that you define within the “Store Hours” topic. These nodes use if/else arguments or logic gates to determine which store location the customer is inquiring about and provide the relevant hours and contact information.

However, it is not always possible to anticipate every question a customer may ask. To address this, Copilot Studio includes a powerful AI-powered feature that utilizes the latest advancements in NLU models. Your copilot can be linked to a public or Bing-indexed website, allowing it to automatically generate responses in a conversational and plain language manner, without the need for the copilot builder to create topics for every possible scenario.

The AI behind your copilot is powered by the Azure OpenAI GPT model, also used by Bing, which can create copilot topics based on a simple description of your needs. Additionally, you can easily modify and update any topic within your copilot by describing the changes you want to make.

Below is a Copilot I have created for a fictional car washing company that hands customers price vouchers for their washing services. I only have a few basic topics, and I need to quickly create one more topic so that my bot can provide relevant information to users and allow them to purchase a car wash.

Using natural language, I can create a copilot that offers the user available car wash services that they can purchase.

Right away, Copilot Studio goes to work and creates a topic that customers can use to learn more about the services we offer. Notice how trigger phrases have been created for me based on the topic description I provided above.

It would be amazing if I could connect a list of services I offer to my Copilot. I can accomplish this with entities. Copilot conversations use natural language understanding (NLU) to determine a user’s intent based on their input. For instance, if a user types “I would like to order a deluxe package,” NLU helps to identify and direct the user to the topic related to services offered, even if the exact phrase isn’t listed as a trigger phrase.

A crucial component of NLU is the identification of entities within a user’s dialogue. Microsoft Copilot Studio includes a set of prebuilt entities, representing some of the most commonly used information in real-world conversations, such as age, colors, numbers, and names. The use of entities allows the copilot to recognize relevant information from the user’s input and store it for future use.

Let's create a custom entity to hold our services and offerings. Notice that I left "Smart Matching" turned on. The Smart matching option enables the bot's understanding of natural language. This can help match misspellings, grammar variations, and words with similar meanings. With smart matching, if the user types "cost free washes" the Copilot will match the phrase to the closest entity item. In this case that would be "Free washes"

Now that the service offerings entity has been created, I can link my car wash services to a question node that ask the user "which service would you like to purchase." when a selection is made, the choice is saved using a variable that we can reference later called "ServiceName"

Finally, the bot will ask the user for the price of their selected service (as listed on the voucher they received from the company), and for any additional notes before the car wash is scheduled. Both responses are stored in variables as well.

The topic concludes by informing the user what service they purchased, how much it cost, and any car wash notes that were provided.

The published Copilot

The now finished Copilot can offer available services from the entity we created when prompted.

When a selection is made, the user can input the price of the service (as listed on their voucher) and any wash notes the company should be aware of. The conversation ends with the Copilot letting the user know which service they are purchasing, the price, and any notes they provided.

This entire process occurs within a single topic. Countless other topics can be generated to build an even more robust and intelligent Copilot.

Next time we will dive further into Copilot Studio and explore security and governance.