Documentation for Chatbot Integration

Objective

The primary objectives for integrating a chatbot are:

• Personalized Experience: Provide users with tailored interactions based on the user’s information and previous action.
• Time-saving: Go directly to the right part of the conversation, and save the time to reexplain the problem.
• Enhanced Engagement: Keep users engaged with interactive and real-time responses.

1. Context on a Web Integration (widget)

• Embed the chatbot directly on your website. Please carefully read the widget integration documentation before this doc.

Chatbot contextualizing can be achieved depending on the use case making the context dynamic or static :

• 1.1. Static context : Used when the conversation has to be contextualized depending on static data that doesn’t change frequently.

  Dialonce handles such context in different ways to cover all possible use cases :

  • 1.1.1 According to page url : According to the page url where your widget is embedded we can run a regex to make the chatbot aware of the page context.

    Example :

    1. You embed the widget on a page containing “toys” in the url and you want the agent aware of that so he will know that the discussion is about toys.

    Process of integration :

    • You select the page or set of page that should give the same context for your agent.
    • You build a regex matching those page(s) url(s).
    • This feature is not directly available on our public builder. Please ask us to map this regex to an entry point in the conversation or to pass this context to the agent.

  • 1.2 Button clicked context : According to the html id of the clicked button on the page to open the chatbot, we are able to map a context that will be passed to the chatbot.

    <button id="unique-id"></button>
    

    Example : The chatbot is embedded on a web page (www.example.com/account) only available to admin users and you want the agent to display “admin” only features. You embed the widget and choose a unique html id for the button that opens the chatbot, for example “admin-only”.

    <button id="admin-only"></button>
    

    We will map “admin-only” to a context that makes the bot aware that he will only discuss with administrators, offering them “admin” features.

    Process :

    • You embed the chatbot on the page and choose a unique id for the opening button.
    • This feature is not directly available on our public builder. Please ask us to map the button id to an entry point in the conversation or to pass this context to the agent.

• 1.2. Dynamic context:

  <aside> ⚠️ THIS FEATURE IS CURRENTLY UNDER DEVELOPMENT, THIS PART MIGHT BE SUBJECT TO CHANGE

  </aside>

  The client will be allowed to send dynamic information ( e.g. client id, account balance…) to the chatbot before the user session starts for a customized experience per session.

  Example :

  After the user logs into the page where the chatbot is embedded, you want it to be aware of his name, surname and userId.

  Process :

  Contact us to discuss which contextualization you want to setup so we can define what informations you have to pass in the window._dialonce.context object.

  1. When the page is loaded in user browser, define window._dialonce.context :

  <aside> ⚠️ window._dialonce.context should be defined quickly before the user clicks on the chatbot button, if it’s not found when the user clicks, the conversation will not be contextualized.

  </aside>

  <aside> ⚠️ window._dialonce.context.data If provided, this object must contain only string keys and string values

  </aside>

  window._dialonce.context = { 
  		intent: "payer_documents", // intention initial, permet d'aller à un endroit précis de l'arbre
  		entities: {"documents":"facture", "produits":"éléctricité"}, // entité (mots clés) inital. Permet d'aller à un endroit précis de l'arbre grâce aux mots clés
  		startMessage: "Je veux avoir ma facture d'éléctricité pretty please", // attention, ce message permet d'overide l'intention et les entités et les remplacer par un message qui va être envoyé au chatbot
  		data: {
  		  first_name:"John", 
  			last_name: "Doe", 
  		},
  		externalRef: "abcd" // Not part of context but will be used 
  												// for tracking purposes
  }
  

  <aside> ⚠️ Notice that the externalRef is mandatory, it will be used to track sessions without personal data and also to provide KPIs if we have several contexts use cases for the same chatbot.

  </aside>

  Description of the fields expected in the context:

  • intent: Represents the user's initial intention, which in this case is "payer_documents", directing the conversation to a specific part of the decision tree.
  • entities: A dictionary of key entities (keywords) that help refine the conversation path. Here, the keyword "facture" (invoice) is used to navigate to the correct part of the decision tree.
  • start_message: A custom user message ("Je veux avoir ma facture d'électricité pretty please") which, if provided, overrides the initial intent and entities to generate a chatbot response based on this message.
  • data: Contains user-specific details such as "first_name" and "last_name" (e.g., John Doe).
  • language: permet de filtre selon une langue pour une architecture multi-langue.
  • brand: permet de filtrer selon la brand désiré pour des architectures multi-dimensionnel de chatbot
  • externalRef: A tracking reference ("abcd") that’s not directly part of the context but used externally for tracking purposes.
  • availableLocales : comme pour availableLocales à la racine de chatbotParams, cet objet permet de définir les langues souhaitées dans un contexte statique uniquement.

2. Context on any custom digital integration (iFrame or webview for example.):

This comes in addition to the instructions for the ICP integration: ICP - Intelligent Contact Page

When needed to transmit context, you will be able to pass context via query parameters to the do-chat url.

Please follow these steps to transmit context: