102 Alexa, show me the steps.

To build an Alexa Skill, it is more than just saying “Alexa, create a skill”.

Template — The Skeleton
Template is like the skeleton of the Skill. It outlines and defines the Skill basic features. To choose a right skeleton, you need to understand what kind of skill you want to build and Alexa supports.

Alexa support 6 types of skills and I classified them into two groups:

Amazon provides 4 pre-built interaction models for 4 different tasks. If you are developing a skill that matches one of the tasks, you can simply choose the pre-built model as a template. It will save you effort and time in model design.

For example, if you decide to create a Flash Briefing Skill to tell user news feed, you only have to feed the skill builder with the Url.

However, if the pre-build models do not match your need, the only choice is to create a custom skill. This will lead you to a different page.

In the Custom Skill page, you will be asked for details such as invocation name, intents, interface settings etc. Wondering what they are? No worries, we will talk about them later!

Interaction Model — The muscles

Interaction model is like muscles in our body. They reach out to achieve what we want to do. In Alexa Skills, it refers to the conversation between user and Alexa. Without this, Alexa cannot understand you nor me.

Host Service — The brain

Host Service is like our brain. It interprets, thinks and responds from requests received. After a request from the interaction model is received, it is analysed in the host service so as to give corresponding response. For example, when user says “Count down for 1 minute.”, host service identifies the action and value for the action.

Alexa allows host service hosted by Amazon Lambda or your web service. Lambda is an easy path as you only need to paste the Lambda ARN (Amazon Resource Number). Meanwhile, if you host it with your web service, you need to set up the URI for the service and make sure you are using the Amazon trusted certificates.

Ready To Go

When all is set, submission and distribution can be done via Alexa Developer Portal. You will be asked for distribution and certification information in the process. Automatic validation and functional tests will also be run before they are submitted for live release. The tests may not be as detailed as your manual testings, but it helps to find high level issues before your users leave you a bad review!

What’s next?

Next, we will discuss architecture details in custom skills. Stay tuned!