Are you ready to develop skills that can manage complex interactions, follow-up questions, and a wide array of inputs? Do you want to help users find the dog of their dreams? Then our tutorial on Alexa Conversations has you covered.
Alexa Conversations facilitates natural voice experiences on Alexa and uses machine learning to enhance your code, bridging the gap between experiences you can build manually and the vast range of possible conversations.
Our tutorial teaches you the fundamentals of voice design as well as the process of developing a custom dialog manager—giving you all the tools and knowledge to build your own highly-responsive skills.
What to expect in the Pet Match tutorial
To get practical experience with Alexa Conversations, the Pet Match tutorial helps you build a skill that recommends a dog to users based on just a few questions. Along the way, you’ll learn to build an interface that accurately predicts deviations in user responses, tracks conversational context, and handles inputs out of sequence.
Here’s what you’ll find at each stage of the five-module tutorial.
- Lab one provides an overview of the complexities involved with voice design, including the challenges of designing a responsive interface. To contextualize Alexa Conversations, you will get an in-depth look at the difference between single-shot and multi-turn interactions. This covers linear interactions, in which every response goes as planned (known as the ‘happy path’), and non-linear interactions, which deviate from the happy path.
- Lab two explains how to train Alexa Conversations in the context of the skill Pet Match. The skill will ultimately provide users with a recommendation based on three parameters: size, temperament, and energy. Yet what happens if a user provides a parameter out of order, or just one of the three? That’s where leveraging Alexa Conversations comes into play.
Before building Pet Match, lab two offers an in-depth introduction to the training data you’ll need later on. This includes five build-time components: dialogs, slots, utterance sets, response templates, and API definitions. Now, you’ve got the necessary background to understand the ‘how’ and ‘why’ as you begin building in lab three.
- Lab three walks through every step to set up the Pet Match skill and illustrates how to train Alexa Conversations. At this stage, we outline how to use the developer console, how to host your skill’s backend resources, and how to modify the response that Alexa speaks to users. By the end of lab three, you will have all the basis to customize your skill and come away with a welcome prompt: “Welcome to pet match. I can find the best dog for you. What are two things you're looking for in a dog?”
- Lab four gives your skill the functionality to collect the desired slots—energy, temperament, and size of the dog—and provide a recommendation. You'll also train it to send those slots to your skill code via the getRecommendation API Definition. Finally, you'll teach your skill how to respond to users with the information our skill code provides back to Alexa Conversations.
- Lab five adds a second API to your dialog—in other words, passing the result of one API to another API. What if your skill recommends a Goldendoodle and a user wants to know more about the breed? How about if they want to modify a detail or an old response? Alexa Conversations allows you to update your dialog and provide a response from your skill code
Now, you’re prepared to add multiple API definitions per dialog and build complex, responsive skills of your own. For an even more in-depth view, follow along with the course’s real-time video guide.
Once you’ve completed the tutorial, continue the conversation with the Alexa Community Slack at alexa.design/slackand chat with fellow skill developers. You can also reach out to the Alexa Developer team on Twitter @alexadevs.
Take the full tutorial and get started building your next skill.