Updates: Check the release notes for new features and product updates.

Dialogflow integration

When an RBM agent communicates with a user, the agent attempts to guide the conversation with suggested replies, suggested responses, and keywords. These tools prompt users to reply in specific ways that the agent understands and can easily respond to.

However, this type of guided conversation can feel unnatural, and if the user responds in a way that the agent isn't designed to recognize, the agent might not understand the user. To handle user responses in normal language, your agent needs to use natural language understanding (NLU) to interpret responses and transform them into something your agent understands.

Dialogflow is a NLU module that processes natural user input, maps it to known intents, and responds with appropriate replies. By integrating Dialogflow with your RBM agent, you can begin a conversation with a single RBM API call, then let your Dialogflow agent handle understanding and responding to your user. Dialogflow handles rich responses and webhook calls into your infrastructure to make the conversation as personal and vibrant as you want to design.

How it works

When your RBM agent uses Dialogflow integration, Dialogflow handles every message between the user and your RBM agent. However, your RBM agent needs to make an API call to start the conversation. This call includes the phone number of the user you want to contact and the first intent you want to trigger in your Dialogflow agent, typically something to start the conversation with the user.

RBM agent makes an API call

The RBM Platform then contacts your Dialogflow agent to trigger the specified intent.

RBM Platform contacts Dialogflow agent

Dialogflow sends a response, which the RBM Platform sends to the user.

Dialogflow agent sends message to user

When the user responds, the RBM Platform sends that response back to Dialogflow, which processes the user's text and returns a response.

User and Dialogflow agent communicate through RBM

The Dialogflow agent and the user continue to respond to each other, through the RBM Platform, until the conversation finishes.

Design considerations

When you use the Dialogflow integration, keep the following considerations in mind:

  • You need to design your agent's conversation and capabilities in Dialogflow.
  • Google Cloud Pub/Sub is disabled for your RBM agent. Your Dialogflow agent handles all user-generated responses and events.
  • The Dialogflow integration doesn't support the create calendar event suggested action.

What's next?

To integrate your RBM agent with Dialogflow, create a Dialogflow agent and enable Dialogflow for your RBM agent.

Once your RBM and Dialogflow agents are integrated, design responses that leverage RBM's suggestions and rich cards, then start a conversation to test what you've built.