Doing something you have never done can be overwhelming, like turning an idea into an Alexa Skill with an aim to hitch your business to the fast paced voice wagon.

Whatever idea and business model you come up with, the process starts with creating a voice user interface (VUI), which allows people to use voice input to control computers, connected devices, and voice assistants like Alexa.

In our last article, we discussed what a voice assistant is, how they work, and an introduction to VUI. Read it here. Also, to have a consistent flow, I am now taking Amazon's line of voice products and platforms to illustrate.

In the last decade, we used the word UI extensively, which generally stood for user interface and the default assumption was to come up with a graphic design of how your app or website would look like. But, in a voice first world, where there is no screen, no touch, no back button and no strong visual cue than just some blinking blue light ring on an Echo device.

Sounds overwhelming enough? Let us simplify.

Designing a voice user interface requires not just diagrams and tools, but utmost empathy to customers' needs, motivations, contextual usage and temporal setting such as time and place where the voice assistant is invoked. For example, asking Alexa to book a cab is different from interacting with Alexa built in a car dashboard to show you directions. In the former scenario you may be at home or work and have a few extra minutes in hand, but in the latter, you are on the move, and can only make limited eye contact or nuanced conversations while driving. This warrants the voice assistants to provide swift responses and avoid asking any unnecessary or obvious questions.

Designing a VUI starts with the larger context of whether a brand goes to Alexa, or Alexa comes to your brand - there's a huge difference in this two.

Firstly, a brand going to Alexa means adding a new 'skill' and essentially teaching Alexa to do new things like answering questions about your business, playing a podcast produced by you or do things in the real world like book a cab or turn the lights on. For example, when you say, 'Alexa, open India Panchang', it means Alexa can offer almanac information like 'tithi' it learnt from a Skill named India Panchang.

To voice enable a connected product like a mobile app or a smart microwave, we use Alexa voice services, which means Alexa comes to your brand. Once integrated, you can summon the voice assistant within any connected products to get things done. You can voice command a banking app to recharge your pre-paid mobile connection or turn on the microwave and set it for two minutes at medium heat, all with just commands in natural language.

What are natural language commands and how to build them? 

Natural language processing (NLP) is the ability of a computer program to understand human language as it is spoken. NLP is a component of artificial intelligence (AI). But, building this for your voice assistant is not as hard as it sounds because the bulk of the hard work is done by Alexa's brain on the cloud - all you have to do is to understand what could be the 'intent' of your customer when they are using your Skill.

For example, on a website you may have a call to action button that says 'accept', but in a voice first world, a customer may say any of the words including, ok, sure, alright, sounds good or agreed, to convey their intent as accept. These words are called utterance in VUI and to build them, you just have to empathize with your prospective user and key them in, while Alexa does the rest of the magic.

Building dialogue sketch and experience flow:

Once you have figured out the purpose and context of your Alexa skill, it is time to build a simple dialogue sketch. A simple role play between two people will suffice, start by assigning one team member to enact as Alexa and another person as a prospective user. Now, start a simple linear question and carry the conversation while avoiding eye contact between the two. Voila, you have a dialogue sketch!

Caveat for the person playing the role of Alexa is that they must behave within the scope of the skill. For example, if you are enacting a banking skill, then you should not answer questions related to airports or book a cab even if the user asks - you must limit to providing responses in banking information and services.
Now that you have a basic set of questions and answers, imagine how you can enhance the experience. What are the on-boarding steps? How can you repair a conversation if the user's intent is unclear or how should Alexa resume if the conversation stopped midway? Figuring out these edge cases, sums up the overall experience flow of your VUI.
Developing VUI is not as hard as designing a mobile app or website, but building a great experience in a voice first sans visual world takes persistent obsessiveness in understanding customer behaviour. Do try for more resources to design a better Alexa Skill.
The writer is co-founder of Agrahyah Technologies, a preferred partner by Amazon in India for building Alexa Skills. He tweets @sreeraman