Press enter or click to view image in full size
I may not have ESP, but I know something about the future: voice-controlled computing is about to take over — and soon, it’ll be more prevalent than beers at a bonfire. Since early adopters get the worm, now’s the time to start coding for this interface.
That in mind, I decided to build my first Amazon Alexa Skill (a.k.a. Alexa’s equivalent to apps). In doing so, I learned a few small things that make a huge difference in the process. To save yourself time and hassle, read the tips below before starting your own build.
Voice-controlled computing is about to take over — and soon, it’ll be more prevalent than beers at a bonfire
The DL on My Alexa Skill
For my trial run of Skill building, I decided to create a conversational interface for thewirecutter.com. Doing so would give me ample material to play with and hey, who doesn’t love learning about gadgets?
I sat down and started coding. Two to three afternoons later — that’s all it took — I had this demo up and running:
Thanks to the great API the Alexa team created, the build was as delightful as a mountain of fresh-baked cookies. Nevertheless, there were a few things I learned that will make my next build go smoother, sleeker, and faster than the Concorde … What’s that? You want to know what I learned? Coming right up:
The 7 Hard Lessons Learned
While, from a technical point of view, building Skills was as easy as slam dunks for Shaquille, the voice interface design took a bit more thought.
From a technical point of view, building Alexa Skills is as easy as slam dunks for Shaquille
After lots of trial and error, I came up with 7 best practices for creating seamless voice interface. Employ them in your Skill building, and you’ll be golden!
- Carefully plan dialog flows in advance, and make sure all cases end in an actionable response.
- Stick to a low number of intents (a.k.a. options Alexa gives in responses). This makes it easier for the user to understand Alexa’s reply.
- Keep commands short.
- Keep Alexa’s answers short.
- Always give users options at the end of each answer.
- Be aware: you can only use up to 50k static variables in sentences (excepting a few generic categories like dates, numbers, first names, and US cities.)
- Ixnay words supercalifragilisticexpialidocious. Use simple words to ensure Alexa understands what the user is saying.
Bonus Thought on Branding
Want to make your Skill next level? Think about branding. Yeah, I know this sounds as logical as smearing peanut butter all over your kitchen table — There’s nothing visual here, after all! — but hear me out.
My friend Lance brought up branding last week while prototyping an Alexa Skill at Singularity University. Just because Skills don’t have the typical branding outlets that traditional interfaces do — colors, icons, and other visual elements — it doesn’t mean Skills can’t be branded. Rather, your brand is defined through voice. Think about it: whether you start a sentence with “Hello Sir,” or “Cha Bra,” sets a tone, and thus a brand!
Go Forth and Build
And just like that — boom! You’re officially armed with all the knowledge you need to build a killer Skill. Now, go out and build the future! And, be sure to keep me updated by leaving a comment below. I’d love to know how it goes.