Learn to use dialog management in Lambda functions

Use Dialog Management in Lambda


Once Dialog Management has been enabled for your skill (by making slots required), at every turn of the conversation, Alexa includes the status of slot collection as the `dialogState` property in the JSON request sent to your skill. You can choose the skill's response based on this property, rather than checking the status of every slot at every turn, reducing the potential complexity of the code. `dialogState` can take any of the following values inside a JSON request:

- `'STARTED'`: a multi-turn conversation* has been initiated
- `'IN_PROGRESS'`: Alexa is in the process of eliciting/confirming slots and confirming intents
- `'COMPLETED'`: Alexa has all slot values and confirmations needed from the user

*A conversation in which Alexa must ask the user follow-up questions to elicit/confirm slots and intents. The conversation on the right is one example.



Dialog State I


With Dialog Management, you can specify prompts and user utterances for _slot elicitation_ and provide prompts for slot and intent _confirmation_. All of this can be done in the <a href="https://developer.amazon.com/alexa/console/ask?&sc_category=Earned&sc_channel=RD&sc_campaign=DialogManagement&sc_publisher=Codecademy&sc_content=Content&sc_detail=Design_Course&sc_funnel=Visit&sc_country=WW&sc_medium=Earned_RD_DialogManagement_Codecademy_Content_Design_Course_Visit_WW_AdvancedDevs&sc_segment=AdvancedDevs" target="_blank" rel="noopener noreferrer">Developer Console</a>. 

These features take care of slot/intent collection/confirmation so that your skill's Lambda function can focus on its core functionality, like recommending videos or booking a flight. If it also had to manage the state of the conversation &mdash; has the user provided the values for all required slots? has the user confirmed the slots? &mdash; it would need a mess of hard-to-maintain logic.

Instead, Alexa keeps track of the status of slot collection (called `dialogState`) and includes it as part of the JSON request sent to your skill. Your Lambda function can check this status and act accordingly: as long as slot collection is incomplete, delegate control back to Alexa so it can prompt the user for more slot values. 

In this lesson we'll learn about the `dialogState` in the JSON request and how to delegate control to Alexa in Lambda functions.




Introduction

At every turn of the conversation your Lambda function can access the dialog state stored in the JSON request. Using the Alexa Node.js SDK, it can be accessed with `this.event.request.dialogState`.In the next few exercises, you'll see how you can use this property to direct the conversation.

Dialog State II


As long as the `dialogState` is not `'COMPLETED'`, we want to delegate control back to Alexa, letting it handle slot collection for our skill. We can do that within our Lambda function by using the statement `this.emit(':delegate')`.

This _directive_, or instruction, tells Alexa to continue handling slot collection using the prompts and utterances we provided in the developer console. Once all required slots/intents have been filled/confirmed, we can execute the main functionality of the intent &mdash; recommend a video!

Going back to our example conversation: `dialogState` is not `'COMPLETED'` until the user has provided a slot value for `videoType` (movie vs. tv show) and `genre` (action, comedy, horror, drama). Until that happens, your function should delegate slot collection to Alexa. After that point all required slot values are collected and `dialogState` is `'COMPLETED'`. Our Lambda function can then use the collected slot values to make a relevant video recommendation.




Delegate Directive

In this lesson we learned how to check the dialog state in our Lambda function and delegate control back to Alexa:
- If `dialogState` is not `'COMPLETED'`, delegate to Alexa using `this.emit(':delegate')`. 
- Else run the main functionality of the intent already provided in the code.
- `'COMPLETED'` means that all required slots/intents have been filled/confirmed. 

For more complex or custom conversations, this method of static prompts and responses set in the Developer Console may not be flexible enough. Sometimes we need to programmatically take control of Dialog Management in our Lambda function. That's exactly what we'll be doing in the lesson <a href="https://www.codecademy.com/courses/learn-alexa-dialog-management/lessons/control-dialog/exercises/intro-control?action=resume_content_item" target="_blank" rel="noopener noreferrer">Control Dialog in Lambda</a>.

