Amazon Alexa Frailty: Syntax Fragmentation in the Skill Ecosystem

Amazon’s Alexa Frailty

Amazon Alexa Frailty: Syntax Fragmentation in the Skill Ecosystem

Amazon Alexa Frailty: Syntax Fragmentation in the Skill Ecosystem

Amazon Alexa has a frailty problem and it’s me. I can’t remember all of the special syntax required to make its skills work. Unlike an app, where I click and am brought into an environment that offers visual clues and instructions, I’m stuck with Alexa either not understanding me or lecturing me. Neither of which is acceptable.

Amazon has failed to create a natural language interface that can be shared seamlessly with the skills produced for the service.

For example, to receive a Shakespearian insult, I have to say:

“Alexa, ask Shakespearean insult to insult me.”  rather than “Alexa, give me a Shakespearean insult,” or even better, “Alexa, insult me like Shakespeare would.”

Alexa clearly needs to invoke an app, and that app has its own verbal command-line prompts that must be followed. Sometimes I feel almost like I need to add a dash-l at the end of phrase to provide the right parameter.

Siri handles third-party integration somewhat more gracefully for things like Apple HomeKit integration. When I ask Siri to turn on the living room light, I simply say, “Turn on living room,” and Siri says, literally, “Your wish is my command, Living Room on.” While a good example, it doesn’t say much about Apple’s Siri integration because its offers few compared with the Alexa skill ecosystem. With Alexa, I must remember which of the many skills I’m talking to and how to talk to them. The most natural integration is with the Insteon system, which works similar to Siri” “Alexa, turn on Bed1,” which Alexa returns with a curt, “OK.”

Many of Alexa’s apps require that they be opened, such as “Alexa, open the bartender.” or invoked like an incantation, “Alexa, ask the bartender to pick a drink for me.” Both put me into “bartender mode” meaning other questions can’t be answered until I leave the app. If I simply ask Alexa to pick a drink for me at the root level, I get “sorry, I don’t know that.” If I ask for a recipe, however, Alexa pretty naturally offers a suggestion, and will read off the ingredients. That similar services require different invocations and syntax means an inconsistent interaction model at minimum, and a failure to understand the need for such a model at the design level.

I understand that skills are actually apps, and apps aren’t just a pool of new capabilities. I expect, however,  that Alexa work as a pool of skills. If I add a new skill, the skills should not create an app wall, but act as an extension of the general knowledge base. Once I add the bartender skill, Alexa should just know about bartending and answer questions just like it does about recipes.

Of course, this breaks the app model. Once an owner adds a skill, they don’t need to add the same skill again. In the app world, having multiple calculators or music apps is the norm. Amazon recently started moving into advertising which will likely result in further balkanization of the Alexa experience rather than improved integration. On the other hand, a more fluid dialog would be able to insert suggestions with more subtly than launching a Procter and Gamble skill in hopes of finding a Tide coupon. Suggesting Bounty paper towels to complement a Pamper order would make sense, especially since Amazon is going to be able to calculate the likelihood that I will need more paper towels given the last time I pushed my Dash button.

As much as Amazon has worked on core integration of Alexa with its services, the ecosystem leaves much to be desired when it comes to integration of third-party services. Amazon needs to make an investment in fluid user experience even if it means breaking existing apps. The frailty of its skill model means that people are not using Alexa as much as they might. And that is a bad omen for Alexa in the long-term. For people to assimilate Alexa into their daily workflow Amazon needs to deliver a much more holistic skill model that doesn’t rely on remembering specialized phrases. In the end, its the stickiness of the entire experience that will determine the real success of Alexa, not that it can just place items in a cart or play Amazon Prime Music. The Amazon Alexa Frailty: to become more than a novelty, Alexa needs to adopt an epistemology, not just an ontology.

More Serious Insights on smart assistants and Alexa read 8 CES 2018 Untrends.

Daniel W. Rasmus

Daniel W. Rasmus, Founder and Principal Analyst of Serious Insights, is an internationally recognized speaker on the future of work and education. He is the author of several books, including Listening to the Future and Management by Design.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.