12 reasons to be developing Alexa Skills right now.

I haven’t had this much fun as a developer in over 20 years. Here are some of the reasons I find working on voice applications — and the Alexa platform in particular — a joy:

1. You’re innovating again It’s fantastic to be creating applications thinking “hey, nobody has ever done that before”. No reference sites or apps to reference, just your imagination. After 20 years of full stack dev, it’s a breath of fresh air.
2. It’s super to be experimenting & learning. Outside of the core Alexa team, few people could call themselves an “expert” at conversational UX right now — we’re all experimenting and learning.
3. Building for an environment, not the screen in your hands. Speak to a device on the other side of the room and make things happen to the environment — lighting, video, audio, temperature. And we’re just getting started with the whole connected ecosystem of Echo devices, Fire TV, and car integrations.
4. Hacking. It’s like creating the first Javascript games for Netscape 2 when you’d create hangman or tetris with some awful hack using a frameset, Javascript timeout, and frame reload since image replacement wasn’t even possible yet. In retrospect, it was crap but at the time it was beautiful. Solving problems until the technology caught up.
5. The community is still small. You can still get answers, people help each other. Though it’s a long way off the quality and cosiness of 90s mailing lists, the “I need this for my homework, please give me the code” contingent are thankfully not there yet.
6. Focus on functionality. No labouring over colours, pixels, animation, the client’s favourite colour, filling whitespace. No wasting time trying to make that page unlike the site your designer ripped off. Just stay focused on the problem space.
7. Your Skills get found and used. Compare being one of 15,000 Alexa Skills to being one of 2.2 million apps in the Appstore, or one of over a billion websites. And end users are looking for new Skills for their devices.
8. The bar is set pretty low. I am just flabbergasted at how few Skills have been developed by companies yet. And of those, many are unrated or are quite dismal. It reminds me of a quote which resonated with me in ’95 from Michael Wolff’s Burn Rate — “if that’s success, I can’t fail”.
9. Easy to test. Write the code — speak to your device sat on the other side of the room — listen to the response — check your screen for errors — repeat. Even your spouse and kids can get in on the testing.
10. NLP and lifelong learning While the Alexa SDK takes care of much of the heavy-lifting using a similar paradigm to the conversational design of chatbots/api.ai , there quickly comes a stage when a slot-based approach becomes unwieldy. As the number of possible access vectors grows, it’s better to roll your own NLP and that’s a fun problem space. However, as somebody once said, “saying that you’re doing NLP is like announcing that you’re going to study medicine”.
11. Marketing hasn’t ruined it yet. Some technical features are genuinely useful — for example, Notifications/Alerts which developers have been clamouring for. However, it’s only a matter of time before these get abused in the name of ‘retention’.
12. It is wonderful to be part of something transformational . Realtime sentiment analysis of spoken voice, long-form voice-to-text messaging/dictation, on-device NLU/machine-learning… it will all come in time.

Of course, there ARE downsides too:

• All that hacking could be for nought. The APIs and core tech will mature, later developers won’t need to spend as much time working around bugs or limitations. Frameworks and tools will allow them to write once for all platforms or without any code at all. That said, there’s a reason most people still hand-roll their own HTML/Javascript/CSS today — there’s still more power to knowing the underlying protocols.
• Limited market demand? Talking about voice today takes me back to 1996, trying to convince companies that this Internet thing was really a big deal. At times it looks like ‘lean days ahead’, waiting for market adoption to reach critical mass outside of the USA or UK. Time to relocate perhaps.

Originally published at www.voiceflow.io on August 11, 2017.