Dustin Coates in Googleassistant

Year in Review: 2017 in Voice-First News for Actions on Google/Google Assistant Developers

Amazon wasn’t the only one having a big year in 2017. Google, too, continued its advance. Still in second place, but a very strong second place. (While I don’t think the situations are at all the same, Google’s been in second place before with Android and seen its patience rewarded.)

API.AI is Now Dialogflow

The biggest superficial change was a new name for API.AI. Now Dialogflow, reflecting that the service focuses primarily on building conversational agents. Of course, you can still build for Google Assistant without Dialogflow, but it’s still most developers’ primary approach.

That wasn’t the only change for the tool. Dialogflow also launched an enterprise edition, added voice recognition, and moved to be under the Google Cloud umbrella. For developers and users, the voice recognition is surely the most important since being under a single roof makes for less latency.

New Devices, New Languages, New Locations

Not long after Amazon announced their new Echo devices, Google had a couple announcements of their own: Google Home Mini and Google Home Max. The Mini is undoubtedly an attempt to take on the Echo Dot. Not a bad idea when you consider that the Echo Dot was the best selling product on Amazon for Christmas 2017. The Max can be seen as an attempt to get a foothold in the “audio quality is important” segment of the population before HomePod comes out. It’s a smart approach, and you wonder why Apple sat out the holiday season this year. (There have been a lot of predictions—my own included—that Apple has proven wrong, so I won’t make any guesses as to how HomePod will sell.)

Google also took an eye toward internationalization in 2017, with support for English, French, German, Japanese, Italian, and Portuguese. Google Home made its way to new countries, too: landing in Canada, Australia, Japan, Germany, and France to join the US and UK. That means that Amazon and Google are nearly at parity with countries, although Amazon has gone to a billion person market in India and a smaller one in New Zealand, while Google is in France by itself.

Google Assistant SDK

Much like Alexa or Android, Google Assistant doesn’t exist to sell specific devices, but to be a platform. The more a family becomes a “Google household,” the less likely they are to switch to Bing or another search engine. Under this model, the Google Home line of devices is just one delivery channel—a POC or a gold-standard, depending on your point of view.

To achieve this, Google launched the Google Assistant SDK, providing a tool for hardware developers to add Assistant to their devices. In what has become a bizarre (in the best possible way) trademark for Google, there was also a cardboard voice kit built on top of the SDK. With the Google Assistant SDK, you can start to see products like the LG Thinq speaker, Bose headsets, and others. The SDK goes beyond just embedding Assistant into a device and controls the device capabilities (the “hello world” would be turning on an LED light, for example). As of late December, it also supports text-based input and output.

Assistant Application Developer Functionality

There was a lot of new functionality for Assistant application developers. Let’s start with how Assistant responds. A new sound library brought over from YouTube gives developers sound effects like “aggressive motorcycle,” “girls’ soccer game,” and “fart toot [sic].” T.S. Eliot fans be sure to check out “human eating a peach” to understand why Prufrock had such a hang up about the whole ordeal.

The sound library goes hand-in-hand with powerful improvements to the SSML functionality that adds par, seq, and media. These are all non-standard tags that give more control over how sounds are added to the SSML. par allows parallel sounds, seq sequential responses, and media options like how many repetitions, fading in or out, etc.

There was news on the updates and notifications front, too. Developers can send both via Assistant, although the functionality is still in developer preview and apps cannot be published with this functionality. Assistant apps can send daily updates, which are sent at the same time every day. For non-regular notifications, apps have push notification capabilities on opted-in intents.

Account linking became easier, or at least more user-friendly, in 2017. With AskForSignIn, developers can now ask for account linking precisely when it is needed. In the past, apps that may need account linking had to ask for permission at the very beginning. A high bar if the app only needed it for a subset of users. For more tentacles into your users’ lives, Assistant now has multi-surface conversations where an interaction can be transferred from one device (e.g. Google Home) to another (e.g. an Android phone). Today, it only works in English and only with Android phones, but paired with the Assistant SDK, you would hope it will expand in 2018.

Google wanted to make developing an app easier in 2017. Perhaps they saw the press numbers of how many Alexa skills there were and saw that, indeed, many of them came from templates. Whatever the reason, Google came out with their own templates. These are quite different, however, and quite ingenious. These templates are built on top of Google Sheets, with no need for code. Interesting for personal apps (like flash cards) or for prototyping more involved apps. For those who go beyond templates and publish an app, the community program offers GCP credits and a free t-shirt.

One area that Assistant is definitively ahead of Alexa is with payments and transactions. I should note that Google is focusing on purchasing offline products through Assistant, while Amazon’s recent transaction announcements focus on in-skill purchasing. Nonetheless, you must credit Google for getting there first. Are you going to build a voice-only business on this transactions API? Probably not, as it looks geared toward getting the big brands there.

I hate to keep coming back to the comparison, but I am thinking about the year-end recap for Alexa and comparing it to this year-end recap for Assistant. There were certainly more high-profile announcements from Amazon than from Google, but the search giant has been doing their best to keep pace. 2018 will be a big year for voice-first in general, and the question for Google is: do they gain ground on Amazon, or do new entries from Apple, Samsung, and even Facebook dethrone them from their AVIS position?