The Alexa Conversations dialog manager for the Alexa Skills Kit is now generally available in the en_US locale. Alexa Conversations is a new AI-driven approach to dialog management that enables you to create skills that customers can interact with in a natural, less constrained way - using the phrases they prefer, in the order they prefer – while freeing you to focus on the highest value parts of your experience. Thousands of developers around the world have now engaged with or published new voice experiences using Alexa Conversations. Brands like iRobot and Amazon, professional Alexa skill builders, and hobbyists all found it faster and simpler to build and enhance dialogs for use cases like catalog selection, guided inquiry, and scheduling.
Alexa Conversations is the first and only deep learning-based dialog manager generally available for voice development. The Alexa Conversations dialog manager works best for published voice developers familiar with basic AI/ML constructs and experienced with Node.js and/or Python programming languages. Visit the authoring tools in the developer console to start building, sample the Pizza Ordering reference template, learn from online educational content and documentation, and explore the Alexa developer forum for support.
Beta participants found Alexa Conversations faster than traditional techniques (once they understood the concepts) for building complex, multi-turn interactions such as helping customers schedule activities or browse a digital catalog. Offloading state management and contextual memory work to Alexa freed them to integrate external APIs or build rich soundscapes for engaging and immersive experiences. Constructive beta feedback produced improved error messages, added dialog cloning, drove new command line interface support, enhanced authoring workflows, and updated the design guide to help you fine-tune experiences to feel more natural. These new experiences are now available on Alexa-enabled devices and in the Alexa Skills Store.
A wide range of developers competed in the Alexa Skills Challenge: Alexa Conversations for prizes. Grand Prize winner John Gillilan said “We created Art Museum, a digital museum experience allowing you to browse the Art Institute of Chicago’s digital collection with Alexa. It truly is a new paradigm for skill building, since building dialog management, context carryover and state management by hand with intents and session attributes is possible but it would have been really hard and flimsy. Outsourcing the state management piece took a huge burden off the development process, and integrating with the Art Institute’s public API opened up an entirely new kind of use case.” Learn more about Art Museum and why John and his partner Katy Boungard donated a portion of their prize money to 8 different Chicago Arts, Media, and Tech charities benefitting women and minorities.
The iRobot Home skill gives customers hands-free control for vacuuming and mopping with connected Roomba® and Braava® robots. Adding the Alexa Conversations dialog manager to the iRobot smart home skill means their customers can create robot cleaning schedules using any number of dialog paths, make changes without starting over, and speak more naturally. iRobot’s developers can now also iterate on the skill faster and with less effort. According to CTO Chris Jones, “Our customers can now schedule cleaning jobs in their own words. And since Alexa manages the dialog, we can improve the experience by retraining the AI instead of re-writing and re-testing skill code. This has fundamentally changed our voice development process and we’re excited by how this speeds up innovation for iRobot customers."
Some of Amazon's own teams have chosen Alexa Conversations to deliver entirely new customer experiences. The “Alexa, What Should I Read Next?” feature helps Amazon customers use Alexa to browse books and find new recommendations. Software Development Manager Rajat Jaggi reported, “The Alexa Conversations dialog manager is the most practical way to create and manage the exponential variety of dialog paths required for this kind of experience. Now, for the first time, customers can use voice to explore book recommendations from many trusted sources in a single place, showing one of the many ways Alexa is becoming more conversational.”
The Ring doorbell camera team added the Alexa Greetings feature to interact with people on the front porch, helping Alexa determine intention and take actions like delivering instructions or taking a message. Alexa Conversations enables Alexa to navigate this free-flowing and ambiguous experience gracefully while streamlining construction and maintenance for the development team. “Building this type of experience is nearly impossible with intent-based dialog, and simple iterations usually take the team months because those dialog models can be so complex and brittle,” noted Amazon Software Development Manager Suranjit Adhikari. “During the beta for this new Ring feature a single engineer re-trained the AI to drive a customer satisfaction increase in just weeks."
Command line access to Alexa Conversations was the most commonly requested feature from the beta program. The new Alexa Conversations Description Language (ACDL) (beta) now allows experienced developers to author Alexa Conversations dialogs in a declarative manner. ACDL comes with type safety, semantic validation for early input error reporting, and introduces modularity and reusability. It works with the Alexa Skills Kit Command Line Interface (CLI) and supports a syntax highlighter extension for Visual Studio Code.
The new Alexa Entities (beta) feature lets you resolve strings in a customer’s utterance to popular entities from Alexa’s Knowledge Graph (including people, places and things), using those entities as an entry point to traverse Alexa’s structured knowledge. This feature makes existing Built-in Slot Types richer and more useful, and reduces the effort in sourcing, acquiring and maintain general knowledge content for your skills. Connections between entities can be used to create more engaging, intelligent and natural dialogues, for example “Add Alias Grace to my reading list”, “Got it. Have you thought about reading The Testaments, also written by Margaret Atwood in 2019?” Alexa Entities can help your skill fetch information about Movies (e.g. films by Quentin Tarantino), Countries (e.g. the population and capital of Belgium), Animals (e.g. the average weight of a hippo) and much more.
Try these new beta features by reading the documentation and visiting the Slack community to engage with product teams directly. Both of these new features support Alexa Conversations as well as traditional slots/intents dialog management.
Visit the Developer Console to enable Alexa Conversations, see an example in the all-new Pizza Ordering template, and publish skills in the en_US locale. Watch the office hours recordings and view special community events to learn Alexa Conversations. Join the Slack community to interact with peers, provide real-time feedback , and stay up on the latest features.