Use AI to Create an Integration

On the integration canvas, an AI chat box is available. Specify what the integration needs to do in the AI chat box and click Send message . See Best Practices.

For example, you can specify to create an integration that receives customer information from Salesforce and updates NetSuite with that customer information if the customer is a VIP customer, sends a notification to the customer after updating NetSuite, and logs a message if the customer is not a VIP customer.

Note:

Creation of an integration using AI is only supported within a project.
AI features are available in select regions. See AI Feature Availability.

The empty integration canvas is shown with the AI chat box where you can specify what the integration needs to do and click the Send message icon.

Here’s what happens after you click Send message

.

The AI determines the nodes to be created for the request specified by the user and provides the list of nodes in a sequential order along with the type of adapter to be used for the trigger/invoke.
The connection needed for the integration is automatically selected from the available connections within the project and a trigger/invoke is created. If there is no connection available within the project, the AI will try to create a draft connection and then create a trigger/invoke using it.
The nodes are created too.
If multiple connections are available within the project, then the AI tries to identify the appropriate connection and then creates the trigger/invoke.
Make sure you preconfigure the necessary connections correctly and specify their usage in the chat box clearly. The AI might automatically configure the trigger/invoke by selecting the operation and business object needed for the trigger/invoke based on the natural language text about what the trigger/invoke needs to do and the configured connections specified for use in the chat box. The automatic configuration of trigger/invoke by AI is designed to deliver the best possible results. It is attempted automatically, however successful configuration cannot be guaranteed in every case.
Note:
The automatic configuration of trigger/invoke is supported only for REST Adapter, Oracle ERP Cloud Adapter, Oracle NetSuite Adapter, Salesforce Adapter, Oracle HCM Cloud Adapter, and Oracle CX Sales and B2B Service Adapter if the following conditions are met:
- REST Adapter: Automatic configuration of trigger/invoke is supported only when configured with the Swagger/OpenAPI catalog.
- Oracle ERP Cloud Adapter, Oracle CX Sales and B2B Service Adapter, and Oracle HCM Cloud Adapter: Automatic configuration of trigger/invoke is supported only for CRUD operations, generic/non-custom business objects, and generic/non-custom business events.
- Oracle NetSuite Adapter: Automatic configuration of invoke is supported only for basic/search operations and single generic/non-custom business objects.
- Salesforce Adapter: Automatic configuration of trigger is supported only on the available platform events for the configured salesforce application. Automatic configuration of invoke is supported only for create, update, or delete operations using generic/non-custom business objects.

The AI creates a skeleton integration.

In the Create/updated integration flow option, click Accept AI generated actions to accept the integration flow created by AI or click Remove all AI generated actions to remove the integration flow created by AI.

If you clicked Remove all AI generated actions , the integration flow created by AI is removed and the AI chat box is displayed again, where you can modify your request or specify a new request indicating what the integration needs to do.

If you clicked Accept AI generated actions , the integration flow created by AI is retained and displayed on the canvas. You can modify this skeleton integration to configure the integration as needed. If needed, you can edit the connections and the triggers/invokes. You can activate the integration after you’ve configured it.

To learn how you can edit an integration using AI, see Use AI to Edit an Integration.

Best Practices

Specify the exact system to be called (for example: Salesforce, NetSuite, or others) so that the trigger/invoke is created with the same adapter.
In order to create the trigger/invoke with correct connections, provide the connection name in the AI chat box (for example: Call service A using connection {connection_name in system}).
To add a fault handler, specify conditional logic too in the AI chat box (for example: Notify the user if any failures occur during the process or Handle failures/errors during the process).

Examples

Example 1

Let's use AI to create an integration flow that creates a sales order. You need to specify what the integration needs to do clearly.

Create a Salesforce connection named SalesforceConnection.
Create a NetSuite connection named NetSuiteConnection.
Create an application integration.
On the integration canvas, an AI chat box is available. Enter the following in the AI chat box:
Create a sales order in NetSuite using NetSuiteConnection on closure of an opportunity in Salesforce (that uses SalesforceConnection). Before creating the sales order, create a customer in NetSuite using NetSuiteConnection if the customer does not exist. Notify the user about errors if there are any failures.
Click Send message .
The AI automatically configures the trigger/invoke by selecting the operation and business object needed for the trigger/invoke based on the natural language text and the configured connections specified for use in the chat box, and creates an integration flow.
- The trigger is created with the Salesforce connection named SalesforceConnection fully configured with:
  - Action Group: Streaming API
  - Streaming Event Type: Subscribe to platform events published on Salesforce.com
  - Event: Opportunity_Closure_Event
  - Custom HTTP Request Header: api_key
- The invoke is created with the NetSuite connection named NetSuiteConnection fully configured with:
  - Processing Mode: Sync
  - Operation Type: Basic
  - Operation: Add
  - Business Object: Customer
- The invoke is created with the NetSuite connection named NetSuiteConnection fully configured with:
  - Processing Mode: Sync
  - Operation Type: Basic
  - Operation: Add
  - Business Object: SalesOrder
- The invoke is created with the NetSuite connection named NetSuiteConnection fully configured with:
  - Processing Mode: Sync
  - Operation Type: Basic
  - Operation: Add
  - Business Object: SalesOrder
In the Create/updated integration flow option, click Accept AI generated actions to accept the integration flow created by AI.
The integration canvas shows the integration flow that you accepted.
Configure the actions as needed.
Activate the integration.

Example 2

Let's use AI to create an integration flow that performs operations such as create, fetch, update, upsert, and delete. You need to specify what the integration needs to do clearly.

Create a Salesforce connection named SalesforceConnection.
Create a REST connection named RestTrigger with an appropriate OpenAPI URL.
Create an application integration.
On the integration canvas, an AI chat box is available. Enter the following in the AI chat box:
Create an integration with a REST trigger using the RestTrigger connection, which receives an account ID, and creates, fetches, updates, upserts, and deletes the account from Salesforce using the same RestTrigger connection. Add a fault handler to log the payload for troubleshooting.
Click Send message .
The AI automatically configures the trigger/invoke by selecting the operation and business object needed for the trigger/invoke based on the natural language text and the configured connections specified for use in the chat box, and creates an integration flow.
- The trigger is created with the RestTrigger connection fully configured with:
  - Resource URI: /accounts
  - Verb: GET
  - Query parameters: AccountId
  - Sample JSON response:
```
{
  "BillingCountry" : "BillingCountry",
  "BillingCity" : "BillingCity",
  "BillingStreet" : "BillingStreet",
  "ShippingPostalCode" : "ShippingPostalCode",
  "ShippingCountry" : "ShippingCountry",
  "BillingPostalCode" : "BillingPostalCode",
  "ShippingCity" : "ShippingCity",
  "ShippingState" : "ShippingState",
  "BillingState" : "BillingState",
  "AccountName" : "AccountName",
  "ShippingStreet" : "ShippingStreet"
}
```
- The invoke is created with the Salesforce connection named SalesforceConnection fully configured with:
  - Action Group: Create, Update Or Delete Information
  - Operation: Create
  - Business Object: Account
- The invoke is created with the Salesforce connection named SalesforceConnection fully configured with:
  - Action Group: Query Information
  - Operation: Retrieve
  - Business Object: Account
- The invoke is created with the Salesforce connection named SalesforceConnection fully configured with:
  - Action Group: Create, Update Or Delete Information
  - Operation: Update
  - Business Object: Account
- The invoke is created with the Salesforce connection named SalesforceConnection fully configured with:
  - Action Group: Create, Update Or Delete Information
  - Operation: Upsert
  - Business Object: Account
- The invoke is created with the Salesforce connection named SalesforceConnection fully configured with:
  - Action Group: Create, Update Or Delete Information
  - Operation: Delete
  - Business Object: Account
In the Create/updated integration flow option, click Accept AI generated actions to accept the integration flow created by AI.
The integration canvas shows the integration flow that you accepted.
Configure the actions as needed.
Activate the integration.

Example 3

Let's use AI to create an integration flow that checks the current weather details of a city.

Create a REST connection named WeatherConnection with an appropriate OpenAPI URL.
Create an application integration.
On the integration canvas, an AI chat box is available. Enter the following in the AI chat box:
Create an integration to check the current weather details of a city with connection WeatherConnection.

Click Send message

.

The AI automatically configures the trigger/invoke by selecting the operation and business object needed for the trigger/invoke based on the natural language text and the configured connections specified for use in the chat box, and creates an integration flow.

The trigger is created with the WeatherConnection REST connection fully configured with:

Resource URI: /current
Verb: GET
Query parameters: This includes parameters (such as cityId, long, and so on).

Example of a response body:

{
    "data": [
        {
            "location": {
                "city": "Raleigh",
                "state": "NC",
                "country": "US",
                "latitude": 38.0,
                "longitude": -78.25,
                "timezone": "America/New_York"
            },
            "time": "2017-03-15 T13:11:00",
            "station": "KU",
            "temperature": 13.9,
            "humidity": 30,
            "pressure": 1013.12,
            "wind": {
                "speed_mps": 5.85,
                "direction": 125,
            },
            "precipitation": 2.0,
            "cloud_coverage": 42,
            "visibilitykm": 10,
            "uv": 6.5,
            "airquality_index": 50
        }
    ],
    "count": 1
}

The invoke is created with the WeatherConnection REST connection fully configured with:
- Resource URI: /current
- Verb: GET
- Query parameters: This includes parameters (such as cityId, long, and so on).

The integration canvas shows the integration flow created by AI.

In the Create/updated integration flow option, click Accept AI generated actions to accept the integration flow created by AI.
The integration canvas shows the integration flow that you accepted.
Configure the actions as needed.
Activate the integration.