Alexa Skills Will Never Take Off

I'm obsessed with Alexa Skills. That's a picture of me, proud owner of an Alexa developer sweater, woot woot! As a developer, Alexa skills are super easy to write, especially with open source tools like the alexa-app npm module at our disposal. Combine these with Zeit.co, my latest obsession (Thanks for introducing me Mike Reinstein!), which allows developers to publish modules/websites to the cloud quickly with ssl cert support -- a requirement for Alexa skills -- and creating an Alexa skill is fast.

That's why every company should have an Alexa skill similar to how most companies are expected to have their own mobile app these days. There really is no excuse not to have an Alexa skill. They don't take much developer effort to create one. You literally map a set of utterances (which are just "hardcoded" phrases) with API calls. If your company has existing API calls and is customer facing, you really should have an Alexa skill. It's an easy way to get in the door with claiming very very simple AI integration. As you could probably guess from my picture, Amazon is encouraging developers to start playing around with the Alexa Skill Kit by giving away free t-shirts and hoodies when a skill is accepted each month. Which is probably why this recent Gizmodo article complains, stating "The Amazon Echo Now Has 10,000 Mostly Useless 'Skills'". Developers are pushing out their newbie skills to get the free gear resulting in lots of random fact type skills. I know getting a free sweat shirt pushed me to finally play... it was on my todo list but who doesn't like free stuff???

The average number of skills echo users routinely interact with is probably ZERO

The article goes on to claim that the author "like many other Echo owners -- only use two or three different skills". But the examples he uses are asking for the weather, setting an alarm, and playing music. Those aren't skills, those are the builtin interactions to Alexa's core software stack. If I were a gambling woman, I'd put money on the average number of skills echo users routinely interact with, third party add-ons to Alexa, being closer to zero.

I have two skills available.

LaundryNFC

LaundryNFC was my first skill accepted by Amazon. I have created NFC tags (pictured above) that you put on your washer and dryer. You can tap your smart phone to the tags and the companion mobile application will send you a push notification when your laundry is done. I ALWAYS forget my laundry and have to run it a second... or third... time. So I created the tags out of necessity and because I wanted to play with writing code for NFC. But sometimes I don't have my phone on me when I'm starting the load, so I created the companion Alexa skill so I could just shout to Alexa to start the timer. Then I still get a push notification on my phone once I link the app to my Alexa account. Give it a try and download the free companion mobile apps, I'd love your feedback.

The Internet of Mysterious Things

My latest children's book, The Internet of Mysterious Things, I self-published and raised funds through Kickstarter and am now crowdfunding through iFundWomen.com. I thought it'd be fun to create an Alexa skill as a companion to the already very nerdy book. So the skill allows parents to ask Alexa for random facts about the Internet of Things. You can also ask Alexa to read the story. This one was a little harder to push through Amazon because of COPA laws. I had to be very clear that this wasn't directed at children and not collecting any personal information before my skill was accepted. There is actually a yes/no question on the skill submission that asks if your skill is intended for children. If you check yes, it's automatically rejected. Read more here: https://developer.amazon.com/public/solutions/alexa/alexa-skills-kit/docs/alexa-skills-kit-policy-testing .

For both of my skills, I found it very easy to get my questions answered from the Amazon Alexa developer evangelists and team. I used Twitter to talk to their team and opened issues on StackOverflow.com that were quickly answered.

However...despite my love for Alexa Skills... they will never take off. Here's why:

They're awkward

The whole reason that Alexa, Siri, and Google Home are popular is that we can "talk" to them like we'd talk to anyone else in the room. You can ask them questions using natural language. But Alexa skills go against that ease-of-use paradigm. To instantiate a skill, the user has to first invoke or startup that specific skill. It's not natural.

For example, in the case of my laundry app, a user would say "Alexa, tell LaundryNFC to start my washer" or "Alexa, ask LaundryNFC how much time is left on my dryer". It's weird. If I were asking a real personal assistant I'd simply say "Alexa, start my washer" and "Alexa, how much time is left on my dryer".

I get why Amazon has it working that way. If they had to map each utterance to thousands of skills to determine the intent to trigger, response times would be slow. However, I do believe Amazon could add some smarts to their logic to understand from the subset of skills the user already has enabled which to search on for the matching. You might argue that if they searched all a user's skills, it could be possible for more than one skill to have the same utterance. Something like "who is the creator of this skill" is a common utterance developers add. In cases of conflicts like this, Amazon could ask the user which skill they were asking about. Maybe something like: "There are a few skills that can help you with that, which skill are you asking about?". Seems like a reasonable solution.

I did notice that there are a subset of utterances that Alexa "hard coded" to suggest enabling a skill. For example, if you asked Alexa to order you a pizza she recommends enabling or invoking (meaning stating "Alexa, Ask <Skill Name>" to "Alexa, Open <Skill Name>") the Pizza Hut or Dominoes apps.

Until Amazon makes some changes to the end user experience behind invoking skills, skills won't get the traction that they could and SHOULD. As for me, I'll keep developing them (because they're fun) in hopes that the software catches up to real human interaction expectations.