Fulfillment overview#

What is fulfillment?#

Dialogflow’s console allows to create simple and static responses for user’s intents in conversations. In order to create more dynamic and complex responses, such as retrieving information from other services, the intent’s fulfillment setting must be enabled and a webhook service must be provided:

When an intent with fulfillment enabled is matched, Dialogflow sends a request to your webhook service with information about the matched intent. Your system can perform any required actions and respond to Dialogflow with information for how to proceed.

—Source: Fulfillment.

A detailed example#

sequenceDiagram autonumber User->>+Interface: User query Interface->>+Dialogflow: detectIntent request Dialogflow->>+Webhook service: WebhookRequest Webhook service->>+External API: API query External API->>-Webhook service: API response Webhook service->>-Dialogflow: WebhookResponse Dialogflow->>-Interface: DetectIntentResponse Interface->>-User: Display message(s)

A representation of how data flows in a conversation between a user and a Dialogflow agent.#

The above diagram is a simplified representation of how data flows in a conversation between a user and a Dialogflow agent through an user interface. In this example, the user’s intent is fulfilled by the agent with the help of a webhook service, allowing to handle more dynamic responses, like calling an external API to fetch some information.

The flow of data in a conversation with fulfillment enabled can be described as follows:

  1. The user types a text into the application’s front-end in order to send a query to the agent.

  2. The input is captured by the application’s back-end, which calls Dialogflow API’s detectIntent` resource, either via the official client or via HTTPS request in the form of a JSON. The request’s body contain a QueryInput object, which holds the user’s query (along with other information).

  3. Dialogflow detects the intent that corresponds to the user’s query and, since the intent in this example has the fulfillment setting enabled, posts a WebhookRequest object to the external webhook service via HTTPS in the form of a JSON. This object has a QueryResult object, which also holds the user’s query and information about the detected intent, such as the corresponding action, detected entities and input or output contexts.

  4. The webhook service uses information from the QueryResult object (along with other data from the WebhookRequest object) in order to determine how the conversation must go. For example, it could trigger some event by setting an EventInput, change the value of a parameter in a Context or generate Message objects using data from external services, such as APIs or databases.

  5. In this example, the webhook service calls an external API in order to fulfill the user’s query.

  6. Then, a WebhookResponse object with the generated response data is returned to Dialogflow.

  7. Dialogflow validates the response, checking for present keys and value types, and returns a DetectIntentResponse object to the interface application.

  8. Finally, the application’s front-end displays the resulting response message(s) to the user.