Intelligence API Features

Rate limiting

The Intelligence API was not built to export RAW data on a large scale but to make the calculated results and insights from process mining available to 3rd party platforms and applications. That’s why the Celonis API enforces rate limiting. This means that only a certain number of requests are allowed per day and a certain number of records can be retrieved in each call. Celonis reserves the right to adjust the rate limits at any time to guarantee high-quality service for all clients.

If a client repeatedly exceeds the rate limits or engages in behavior that is deemed to be suspicious, Celonis reserves the right to temporarily or permanently limit or suspend access to the API for that client.

When a client exceeds the number of requests per day, the Celonis API will return a 429 response (too many requests) including an HTTP header (x-ratelimit-reset) which indicates the time (in seconds) that the client needs to wait before a new request can be processed.

The following HTTP headers are also returned as part of each call:

  • x-ratelimit-limit : Represents the max number of requests that the client can perform in the current time window.
  • x-ratelimit-remaining : Number of remaining requests in the current time window.

Currently, the API has the following default limits:

Table 1. Default rate limits

Limit Default Values
Max number of requests/day 6000 requests/day
Max number of requests/second 20 requests/second
Max number of allowed fields per request in the Celonis Platform Knowledge Model 200 fields/request
Max number of records per request returned when calling the /data endpoint 50 records/request
Total maximum number of records that can be retrieved through the /data endpoint. First 1000 records per filtered/sorted table

Pagination

All endpoints that return results in the form of lists provide the ability to paginate their results by leveraging page-based pagination. This can be configured via two parameters:

  • page : the page number to be retrieved starting from 0 . As the maximum number of records that can be retrieved through the /data endpoint is 1000 records , the maximum page number is the one that contains the 1000th record based on the pageSize .
  • pageSize : the number of items per page to be returned. The default value is 50 meaning that the maximum number of items that can be retrieved in each call is 50 records.

Sorting

Data Endpoint

The /data endpoint provides the ability to allow sorting by using the sort query parameter:

  • sort : comma separated list of fields that are being preceded by "+" to display in ascending order, or "-" for descending order. For example: sort=+opportunity_value,-opportunity_id

Note that if no +/- prefix is provided, the results will be sorted in ascending order (+) by default.

Note also, the fields used for sorting must exist for the requested record.

Schema Endpoints

Schema endpoints also provide the ability to sort their results by leveraging the sort query parameter as follows:

  • Multi-field sorting is not allowed, i.e. it Is not possible to sort by more than one field.
  • Fields in the sort query param should be prefixed with a ‘+’ for ascending order or ‘-’ for descending order.
  • If no +/- prefix is provided, the results will be sorted in ascending order (+).
  • The results will be sorted as case insensitive.

Some examples are:

  • Get all Knowledge Models: /knowledge-models?sort=+id
  • Get all Filters for a Knowledge Model: /knowledge-models/{km_id}/filters?sort=-name
  • Get all Records for a Knowledge Model: /knowledge-models/{km_id}/records?sort=id
  • Get all Fields (attributes, augmented attributes and flags) for a Record of a Knowledge Model: /knowledge-models/{km_id}/records/{record_id}?sort=-id

Fields Selection

The /data endpoint allows API consumers to retrieve data for the following record objects:

  • Attributes
  • Augmented Attributes In order to do so, the ids of the objects to be retrieved must be listed as part of the fields query parameter:
  • fields: a comma-separated list of attribute, augmented attribute or flag ids for a given knowledge model record.

Note that:

  • fields is a mandatory query parameter.
  • The maximum number of objects per record that can be retrieved through the API Data endpoint is 200.
  • fields query parameter is case-sensitive. The object ids must match the object ids returned by the record schema endpoint ( /knowledge-models/{km_id}/records/{record_id} ).
  • Each object id must be separated by a comma or, alternatively, use the fields query parameter several times.

Examples:

Copy
Copied
fields=ACTIVE,Category,APPROVAL or fields=ACTIVE&fields=Category&fields=APPROVAL 

Filtering

Filtering allows API consumers to only retrieve the desired records and not the whole data set. There are two ways to filter the data returned by the /data endpoint:

  1. Apply filters pre-defined in the Knowledge Model by using the filters query parameter.
    Filters: a comma-separated list of filter ids.

    Note that:
    • The Celonis Celonis Platform Knowledge Model allows filters to be defined using the FILTER PQL operator as explained in the Celonis Platform documentation .
    • You can apply multiple filters by providing multiple filter ids separated by commas or use the filters parameter several times.
    • Remember this parameter is case sensitive. The filter ids must match the one returned by the filter schema endpoint ( /knowledge-models/{km_id}/filters ).

Examples:

Copy
Copied
filters=not_null_orders,delayed_orders or filters=not_null_orders&filters=delayed_orders
  1. Filter expression using the filterExpr query parameter.
    This parameter allows filtering of record data using a provided value of a field through an expression following:
    1. First pattern field operator value where:
    2. field is a filterable attribute, augmented attribute or flag of a record. This value is case-sensitive and should be equal to the attribute id.
    3. operator is a comparison operator. For this pattern, it is currently supported by following operators:
      • eq(=), gt(>), lt(<), ne(!=), ge(>=), le(<=), is(IS) and in(IN)
    4. value is a constant, such as a Salesforce Opportunity ID or a number/text.
    5. Second pattern operator(field,value) where:
    6. operator is a string search operator,similar to the LIKE in the SQL clause. For this pattern, it is currently supported by following operators:
      • startswith, endswith and contains

    It is also possible to combine filters with logical operators and, or, not and parenthesis.

  2. Example of the first pattern
    Copy
    Copied
    filterExpr=field1 eq value,field2 ne value
  3. Example of the second pattern
    Copy
    Copied
    filterExpr=startswith(field1,value),contains(field2,value)
  4. Example with first pattern, second pattern and or
    Copy
    Copied
    filterExpr=startswith(field1,value) or filed3 eq value,contains(field2,value) or field3 eq value
  5. Example with not
    Copy
    Copied
    filterExpr=not startswith(field1,value) or filed3 eq value,not (contains(field2,value) or field3 eq value)

It’s important to highlight that not all fields of a record are filterable. For example, if you define an attribute as a KPI in the Knowledge Model, this attribute cannot be filtered. You can identify which fields are filterable by checking the headers array in the returned data endpoint JSON, where all fields’ information is shown, including the filterable field.
Example:

Copy
Copied
{
    "page": 0,
    "pageSize": 100,
    "total": 1,
    "sort": [],
    "content": {
        "headers": [
            {
                "id": "description",
                "name": "Description",
                "type": "string",
                "aggregation": false,
        ===>    "filterable": true,  
                "sortable": true
            }
            {
                "id": "count",
                "name": "Count",
                "type": "integer",
                "aggregation": true,
        ===>    "filterable": false,
                "sortable": true
            }
        ],
        "data": [
            {
                "description": "total"
                "count": 50
            }
        ]
    }
}

You can apply several filters separated by commas or use the filterExpr parameter multiple times separated by “&”.

Examples:

filterExpr=opportunity_id eq AA00022475H, opportunity_value gt 50000

Copy
Copied
filterExpr=opportunity_id eq AA00022475H& 
filterExpr=opportunity_value gt 50000

Subscription to Trigger API Features

Team Rate limiting

Subscription to Trigger is enforcing rate limiting. This means for a team, only a certain number of subscriptions can be created. In addition, no matter how many subscriptions are created, there is a maximum number of events that can be emitted from the API to the third parties consumers.

If you reach the maximum number of subscriptions, you'll need to delete an existing subscription in order to create a new one.

If you reach the maximum number of events emitted from the API, the rest of data produced by the Celonis Platform will be discarded.

If you are approaching your daily quota, you'll be informed via emails to your admin account. The first email will be sent when 80% of the quota is exceeded, letting your admin account know that you're over that percentage. The second email will be sent as soon as you exceed 100% of the quota.

Currently, the API has the following default limits:

Table 1. Default rate limits:

Limit Default Values
Max number of subscription per Team 10
Max number of daily events per Team 100.000

Creating a new Subscription

To create a new subscription for an Celonis Platform Trigger you need to consume the next endpoint.

Copy
Copied
POST
/knowledge-models/{{km-id}}/triggers/{{trigger-id}}/subscriptions

{
    "name": "new subscription with headers",
    "callbackInfo": {
        "uri": "https://my-awesome-webhook.com",
        "protocol":"HTTPS",
        "headers": {
            "Authorization" : "my bearer token"
        }
    }
}

km-id refers to the knowledge model that is existing at the Celonis Platform. trigger-id refers to the trigger that is existing at the Celonis Platform.

All the attributes (with the exception of the headers) of the payload are mandatory.

name must be shorter than 1024. callbackInfo.uri uri used for the API to forward events emitted from Celonis Platform. callbackInfo.protocol the protocol used to forward events emitted from the Celonis Platform. It must fit with the callbackInfo.uri attribute. Otherwise it will fail into a Bad Request. callbackInfo.headers map of key/value regarding all the headers that will be attached from the API when the event is forwarded to the consumer.

Updating an existing Subscription

To update an existing subscription, you need to consume the next endpoint.

Copy
Copied
PUT
/subscriptions/{{subscription-id}}

{
    "name": "updated subscription with headers",
    "callbackInfo": {
        "uri": "https://my-awesome-webhook.com",
        "protocol":"HTTPS",
        "headers": {
            "Authorization" : "my bearer token"
        }
    }
}

subscription-id refers to the subscription to update. It must exist in the API.

All the attributes of the payload (with the exception of the headers) are mandatory.

name the subscription name, which must be must be shorter than 1024 characters. callbackInfo.uri uri used for the API to forward events emitted from Celonis Platform. callbackInfo.protocol the protocol used to forward events emitted from the Celonis Platform. It must fit with the callbackInfo.uri attribute. Otherwise it will fail into a Bad Request. callbackInfo.headers map of key/value regarding all the headers that will be attached from the API when the event is forwarded to the consumer.

In case of updating a Subscription which was failing, it will be set back to ACTIVE status.

Pausing a Subscription

To pause an active subscription, you need to consume the next endpoint.

Copy
Copied
PATCH
/subscriptions/{{subscription-id}}/pause

subscription-id refers to the subscription to pause. It must exist in the API.

Once the subscription is paused, it will not forward more events from Celonis Platform, those events aren't lost, they will be send to the consumer once the subscription is resumed.

Resuming a Subscription

To resume a paused subscription, you need to consume the next endpoint.

Copy
Copied
PATCH
/subscriptions/{{subscription-id}}/resume

subscription-id refers to the subscription to resume. It must exist in the API.

Once the subscription is resumed, it will forward new events emitted from Celonis Platform, as well as the data that was pending when the subscription was paused.

Unsubscribing

To remove an existing subscription you need to consume the next endpoint.

Copy
Copied
DELETE
/subscriptions/{{subscription-id}}

subscription-id refers to the subscription to unsubscribe. It must exist in the API.

Once this action is executed, the subscription will be no longer visible at Celonis Platform, and all its data is removed from the API.

Manual extraction (reconcile endpoint)

If the subscription is failing, in the meanwhile the subscription is fixed, is possible to manually extract the data from Celonis Platform in short batches. This is what is called the Reconciliation.

To manually extract data, you need to consume the next endpoint.

Copy
Copied
PATCH
/subscriptions/{{subscription-id}}/events/next-page-from-parking-lot

subscription-id refers to the subscription to reconcile. It must exist in the API.

Once this action is executed, the next 50 events will be taken. Bear in mind this action will remove those events from the system.

Copy
Copied
{
    "pageSize": 2,
    "total": 0,
    "content": [
        {
            "subscriptionId": "XXX-YYY-ZZZZ",
            "triggerId": "97185b79-d6e9-4425-bbee-225627533149",
            "signal": "{\"id\":\"a8ffef1d-aba0-4520-ae1a-02cecc7b7e4e\",\"recordIdentifier\":\"800A3000531000\",\"typedRecordIdentifier\":\"800A3000531000\",\"attributes\":[{\"attributeId\":\"ATTRIBUTE_1\",\"signalId\":\"a8ffef1d-aba0-4520-ae1a-02cecc7b7e4e\",\"value\":null,\"columnType\":\"STRING\"}],\"creationDate\":1699010395821}"
        },
        {
            "subscriptionId": "XXX-YYY-ZZZZ",
            "triggerId": "97185b79-d6e9-4425-bbee-225627533149",
            "signal": "{\"id\":\"8cf03551-471c-41f0-a85c-e88646eef354\",\"recordIdentifier\":\"80000000130071000\",\"typedRecordIdentifier\":\"80000000130071000\",\"attributes\":[{\"attributeId\":\"ATTRIBUTE_1\",\"signalId\":\"8cf03551-471c-41f0-a85c-e88646eef354\",\"value\":\"D1\",\"columnType\":\"STRING\"}],\"creationDate\":1699010395821}"
        }
    ]
}

pageSize refers to the events returned in the page. total refers to the total events still at the API. content events as raw JSON.

Untransferred data

Untransferred data means the data unable to forward to the consumer but saved at the API to retry later. The data can be saved at the API under two circunstances:

  • The subscription started to fail and is still within the time window to sort it out (after 8 hours the service is suspended).
  • The subscription is paused.

Untransferred data can be manually extracted using the reconcile endpoint. But for active subscriptions with data remaining in the API, every 10 minutes our API will retry to forward the data to the consumer.

Finally, the data is kept at the API for 7 days, after the period of time all the pending data will be discarded.