Updated 1 month ago by Matheus Fernandes

A chatbot is a platform for content delivery and end-user interaction. It's related to the Learning Experience step of our Framework and is mainly used for distributed team training. It can be embedded in the web environment, app and external pages (depending on technical implementation by the customer)

This channel works to improve communication with users, scalable personalization of content delivery, and continuous user development. Can be used actively or reactively: Actively when the platform administrator sends messages, encouraging engagement and behaviours; reactively when the user initiates the interaction.


The chatbot's main features are flows and notifications. Let's understand how they work:


Flows are the "conversations" that the chatbot establishes with the end-user, they are activated via triggers which can be a keyword, dates (every Monday at 10:00) or events (when the user first opens the chatbot, for example).

For the keyword trigger, we use Dialogflow, Google's tool that allows reading in natural language, i.e. interpret the different variations of the keyword used.

There are also some types of flows that we need to understand to build the user experience:

  • FAQ: Simple Question and Answer Flow. The trigger of this flow is keyword, remembering that the use of Dialogflow allows natural language interpretation of the keywords used. They may return the answer as:
    • Simple text;
    • Text + Content;
    • Text + link.

It is important to comply with these response restrictions for a proper bot operation. Take a look at the FAQ flow creation template, I think you'll have a better understanding ;).

In the flow type FAQ: Text + Content, you cannot return text-type content.

It works well if you have a wide range of products or high complexity of operation.

  • Chef: This is a flow that returns the chef's recipe content recommendations delivered in the carousel format. The recipe rules are the same as for the web environment, you can also determine how many contents to recommend. The user can still request for more content and, when there are no more recommendations left, the chatbot returns a message stating that all content has already been consumed.
  • Survey: These are flows in which we ask questions to the user, which can be a flow or just a question. The possible responses must be predetermined and can direct the flow of interaction as in a decision tree. Currently, the user responses are not stored so, if necessary, you need to create a special project for integration with a repository for these responses.
  • Custom: Flows that depend on development, the most common examples are: Receive information from an external database (send a seller's goal) and flows that mix various types of flows.

These are messages sent by the chatbot actively. They must answer in the same format as the flows, that is, you cannot send in a different format than the flow models of type FAQ, for example.

Hope this article has helped you!

How did we do?

Powered by HelpDocs (opens in a new tab)