Skip to main content

Rest API

Reference​

Changelog​

2024-02-05​

  • Added monthlyNumberOfProductionWorkflowMinutesUsed, monthlyNumberOfUserMinutesUsed to /organizations
  • Added productionWorkflowMinutes, userMinutes to /plans
  • Added startPage to preprocessConfig in /models and /predictions
  • Added nextPage to POST /predictions output indicating the next page in the document available for predictions, if there are more pages in the document that have not yet been processed.

2023-12-14​

  • Added status, monthlyNumberOfExecutionsCreated and monthlyNumberOfExecutionsAllowed to /workflows. Workflows can either have status development or production. A workflow with status=development will have a low limit on how many executions can be started each month.
  • Added monthlyNumberOfPagePredictionsUsed, monthlyNumberOfPagePredictionsAllowed and monthlyNumberOfProductionWorkflowsUsed to /organizations
  • Now counting number of pages processed each month to determine API limits for the POST /predictions endpoint. Users already subscribed to a paid plan that counts field predictions will still be billed per field prediction. They can choose to subscribe to the new plan which counts pages instead.

2023-12-06​

  • Added statistics to /models/:id. statistics contains aggregated automation percentages for the fields in the model's fieldConfig from the last number of days, defaulting to last 7 days.
  • Added optional statisticsLastNDays to GET /models/:id. Specify a number of days between 1 and 30 to get statistics from.

2023-12-05​

  • Added optional outputFormat to postprocessConfig in POST /models, PATCH /models/:id and POST /predictions. Use this option to specify which structure the predictions will have. The allowed values for outputFormat is currentlyv1 and v2
  • Added optional status to PATCH /workflows/:id/executions/:id. Use this option to change status of execution from succeeded to completed or vice versa.
  • Parameter nextTransitionId is now optional in PATCH /workflows/:id/executions/:id
  • Added GET /appClients/:id.

2023-11-17​

  • It is now possible to set up an email address which can be used to execute workflows. JPEG, PNG, TIFF and PDF attachments in the email will each start en execution of the specific workflow. The configuration requires a secretId which should contain API credentials to invoke POST /workflows/:id/executions
  • Added optional emailConfig to POST /workflows
  • Added optional emailConfig to PATCH /workflows/:id

2023-10-25​

  • Added GET /models/:id/dataBundles/:id
  • Added GET /models/:id/trainings/:id

2023-10-02​

  • Added new endpoint /datasets/:id/transformations. Transformations are operations performed on all documents in a dataset. Currently, the only supported transformation is remove-duplicates.
  • Added GET /datasets/:id/transformations
  • Added POST /datasets/:id/transformations
  • Added DELETE /datasets/:id/transformations/:id
  • Added optional pages to groundTruth for /documents. Specifying pages gives you better results when training models.

2023-09-14​

  • Added optional query parameter modelId to GET /predictions
  • Added new permission action delegate in /roles that specifies which roles can be delegated to other resources such as /users and /appClients. Trying to assign roles to /users and /appClients without the delegate permission action will result in a 403 Forbidden

2023-09-07​

  • Added optional metadata to /workflows
  • Removed deprecated avatar, name and email from /users

2023-08-25​

  • Added new endpoint /roles. Roles can be attached to /users and /appClients to limit which API resources they can access.
  • Added GET /roles
  • Added GET /roles/:id
  • Added roleIds to /users
  • Added roleIds to /appClients

2023-06-27​

  • Added fileserver endpoint that can be utilized to upload larger files than 4.5MB and get thumbnails and resized images of documents. The fileserver URL is contained in the response body of /documents and accepts GET and PUT operations.
  • Added fileUrl to /documents. Use this URL to access the file using your Authorization token.
  • content and contentType is now optional in POST /documents. Instead of providing content and contentType, you may upload the document to fileUrl directly without base64 encoding. The maximum document size you can upload to the fileserver is 64MB.

2023-06-14​

  • Added optional preprocessConfig to POST /predictions. autoRotate, imageQuality, maxPages and rotation should now be specified as part of preprocessConfig instead and will be deprecated June 14th 2024.
  • Added optional pages to preprocessConfig to specify which specific pages to process when invoking POST /predictions. pages is 0-indexed and supports negative indices for reverse indexing such that index 0 translates to the first page of the document and -1 to the last page of the document.
  • All individual parameters in preprocessConfig are now optional and may be specified as needed.

2023-06-07​

  • Added new possible status running-final-evaluation for /models/:id/trainings which takes place after running and before succeeded

2023-03-23​

  • Added preprocessConfig and postprocessConfig to /predictions
  • Removed deprecated width and height from /models

2023-03-02​

  • Added signin with Google.
  • Added GET /profiles/:id, GET /profiles/me.
  • Added PATCH /profiles/me.
  • Added POST /organizations.
  • Added GET /organizations.
  • Deprecated avatar, name and email in /users, they will be removed May 20th 2023. Use /profiles instead
  • Removed POST /signup.

2023-02-27​

  • Added optional rotation to POST /predictions

2023-02-10​

  • Added page to predictions in /predictions when page information is available.
  • Added owner query parameter to GET /models. Use this feature to list pre-trained models. GET /models?owner=las:organization:cradl
  • Added support for creating predictions using a pre-trained model. Specify the organization that owns the model to POST /predictions using the composite modelId form (e.g. las:organization:cradl/las:model:invoice)

2022-12-20​

  • Added warmStartConfig to POST /models/:id/trainings. Use this parameter to specify which trainingId to continue training from.
  • Added support for nested output for /predictions.

2022-10-13​

  • Removed deprecated timestamp from /predictions
  • Deprecated height, width in /models. They will be deleted on December 12th, 2022.

2022-09-28​

  • Added parameters to PATCH /transitions/:id. Use parameters to update transitions with transitionType=docker or transitionType=manual specific parameters.
  • Added parameters.cpu, parameters.memory, parameters.imageUrl, parameters.secretId, parameters.environment and parameters.environmentSecrets to PATCH /transitions/:id for transitionType=docker.
  • Added parameters.assets to PATCH /transitions/:id for transitionType=manual.
  • Deprecated environment, environmentSecrets and assets in PATCH /transitions/:id and will be removed on December 16th. Use parameters.environment, parameters.environmentSecrets and parameters.assets instead.
  • Added query parameter fromStartTime and toStartTime to GET /workflows/:id/executions.
  • Fixed a bug causing GET /documents/:id to fail with status code 500.
  • Fixed a bug causing POST /predictions to fail with status code 500.

2022-09-27​

  • Fixed a bug causing POST /predictions to fail with TIFF images containing 2 or more pages.
  • Fixed a bug causing community /datasets and /models to not be added to new users. Existing users that were affected by this bug must contact support to have them added to their organization.
  • Added additional cpu (512, 1024) and memory (4096, 8192) options for docker /transitions.

2022-09-15​

  • Added query parameter documentId to GET /documents
  • Fixed a bug causing /models/:id/dataBundles to fail with status failed when certain combinations of ground_truth value types and model.fieldConfig were present.
  • Fixed a bug causing /models/:id/dataBundles to show validity=1.0 when no ground_truth values were successfully parsed.
  • Fixed a bug causing /models/:id/dataBundles to fail with status failed when no documents with valid ground_truth values were present.

2022-09-13​

  • Several improvements have been made to summary in /models/:id/dataBundles. You can now see which /documents are duplicates, which groundTruth values are duplicates, the distribution of characters and lengths of your data. You can also see which documents had errors in them while processing the data and which labels in the groundTruth that were erroneous. For a complete list of error codes see list of error codes
  • Removed deprecated types all, alphanum, alphanumext, letter, number and phone in fieldConfig in /models.
  • Added type numeric to fieldConfig in /models
  • Added contentMD5 to /documents
  • Added retentionInDays to /models/:id/dataBundles. This change reflects that PII can occur in data bundles and should thus be removed in accordance with the DPA.

2022-06-24​

  • Added evaluation to /models/:id/trainings

2022-05-31​

  • groundTruthSummary in /datasets is now also counting nested values. This change will not affect existing counts.

2022-05-27​

  • Added optional postprocessConfig to /models

2022-05-19​

  • Updates to the enum type in fieldConfig:
    • Enumerations must be unique and from 1 to 512 characters long
    • Maximum number of enumerations allowed is 500
    • Added support for more characters in enumerations

2022-05-10​

  • You may now set trainingId to null in PATCH /models/:id to make the model inactive

2022-04-22​

  • Added DELETE /paymentMethods/:id
  • Added GET /paymentMethods/:id
  • Added GET /paymentMethods
  • Added PATCH /paymentMethods/:id
  • Added POST /paymentMethods
  • Added paymentMethodId to PATCH /organizations/:id
  • Fixed bug causing /models/:id/dataBundles to fail with status=failed if /datasets provided have 0 examples of any label defined in fieldConfig in /models

2022-04-12​

  • Added sortBy query parameter to GET /documents
  • Added order query parameter to GET /documents
  • Providing null as groundTruth to PATCH /documents/:id is now supported
  • Added sortBy query parameter to GET /predictions
  • Added order query parameter to GET /predictions
  • Added createdBy to /predictions
  • Added createdTime to /predictions
  • timestamp in /predictions is deprecated and will be removed after October 11th, 2022
  • Added planId to PATCH /organizations/:id

2022-04-05​

  • You can now use /models with status=inactive in POST /predictions if you provide a trainingId with status=succeeded.

2022-03-31​

  • Increased limit of maxLength for fields specified in fieldConfig in /models to 512.
  • Fixed a bug causing users to not be added to organization in POST /users.
  • Added optional metadata to /users
  • Added optional trainingId to /predictions response

2022-03-16​

  • width and height is no longer required in POST /models and will default to 801 and 1281.
  • maxLength is no longer required for fields specified in fieldConfig in /models.
  • Added type enum to fields specified in fieldConfig in /models. Specify valid enum values by using the key enum for fields with enum type. See below for example of how to do this.
  • Added monthlyNumberOfActiveModelsUsed, monthlyNumberOfFieldPredictionsUsed and monthlyNumberOfGpuHoursUsed to /organizations
  • Fixed a bug causing creation of summary in /models/:id/dataBundles to fail when using number values in JSON for fields of type digits.

Example of valid fieldConfig after the changes:

{
  "dueDate": {
    "type": "date",
    "description": "Due date on invoice"
  },
  "totalAmount": {
    "type": "amount",
    "maxLength": 12
  },
  "category": {
    "type": "enum",
    "enum": ["TRANSPORTATION", "ACCOMMODATION", "OTHER"]
  }
}

2022-03-09​

  • Added trainingId to POST /predictions. You can now make predictions with a trainingId to test out new trainings.
  • Added trainingId to /models. The trainingId for a model specifies which training is the one that's currently being used when making predictions.
  • Added trainingId to PATCH /models/:id

2022-02-02​

  • Added metadata to /datasets
  • Added metadata to /documents
  • Added metadata to /models
  • Added metadata to /models/:id/trainings

2022-01-20​

  • Now accepting more types for value in groundTruth in /documents. Additionally supported types are number, bool, empty string and nested lists. See the full OAS spec for details.

2022-01-06​

  • Added name and description to POST /documents
  • Added name and description to PATCH /documents/:id

2022-01-05​

  • Added retentionInDays to PATCH /datasets/:id. retentionInDays for /documents already associated with datasets will not be affected by changing retentionInDays of datasets.
  • Added containsPersonallyIdentifiableInformation to PATCH /datasets/:id

2021-12-10​

  • Removed status 'training' in /models. Status 'training' have been superseded by endpoint /models/:id/trainings. /models with status 'training' have had their status to 'active' or 'inactive'
  • Added numberOfRunningTrainings to /models
  • Renamed status 'processing' to 'running' in /models/:id/dataBundles
  • Renamed status 'ready' to 'succeeded' in /models/:id/dataBundles
  • Renamed status 'training' to 'running' in /models/:id/trainings
  • Renamed status 'completed' to 'succeeded' in /models/:id/trainings
  • Renamed status 'aborted' to 'cancelled' in /models/:id/trainings
  • Added email notifications for /models/:id/trainings status change. Email notification will be sent to organization owner and user who initiated the training
  • Added trainingsCreated to monthlyUsageSummary in /organizations
  • Added PATCH /models/:id/trainings/:id
  • Fixed bug causing /workflows/:id/executions to be started twice in some rare cases

2021-12-01​

  • Added /plans endpoint. Use this endpoint to see pricing details for different plans
  • Added GET /plans
  • Added GET /plans/:id
  • Added /models/:id/trainings endpoint. Use this endpoint to initiate model training jobs
  • Added GET /models/:id/trainings
  • Added POST /models/:id/trainings

2021-11-04​

  • Removed restrictions on updating fieldConfig in /models, you may update fieldConfig regardless of model status
  • Added postprocessConfig to POST /predictions. Currently supporting two strategies: BEST_FIRST and BEST_N_PAGES. BEST_FIRST returns predictions from the best page and may skip evaluating later pages if good predictions have already been found. BEST_N_PAGES returns predictions from the best N pages
  • Added documentRetentionInDays to /organizations. This value will correspond to the retention of PII as specified in the DPA
  • Fixed a bug causing monthlyNumberOfDataBundlesCreated in /organizations to not be reset each month
  • Fixed a bug in data bundle creation causing documents with null values in groundTruth to be ignored

2021-10-12​

  • Deleted /batches endpoint. Use /datasets instead
  • Added free signup! When signing up you'll get access to the community models. Go to https://app.cradl.ai/signup
  • Removed outdated CORS allowed headers
  • Added digits and string types to /models fieldConfig
  • Added size limit on input to POST /workflows/:id/executions
  • Updated email for signup, invite and verification code

2021-09-23​

  • Added retentionInDays to PATCH /documents/:id

2021-09-22​

  • Added groundTruthSummary to /datasets

2021-09-17​

  • Fixed a bug preventing resources to be deleted sometimes when calling DELETE /documents/:id or DELETE /assets/:id
  • Fixed a bug causing negative values to be returned for several fields (e.g. numberOfDocuments, numberOfDataBundles)

2021-09-08​

  • Updated Login UI
  • API key is no longer needed

2021-08-18​

  • Fixed a bug preventing a transition execution from starting for transitions with the docker type
  • Added GET /datasets/:id
  • Extended the period in which temporary credentials are valid upon first time invitation to Typenode or Flyt from 30 days to 90 days
  • Listing endpoints like DELETE /documents and GET /documents should now respond faster
  • Added more descriptive error message for 404 responses
  • Added createdBy, updatedTime, updatedBy to /appClients
  • Added createdTime, createdBy, updatedTime, updatedBy to /users
  • Added createdTime, createdBy, updatedTime, updatedBy to /assets
  • Added createdTime, createdBy, updatedTime, updatedBy to /documents
  • Added createdBy, updatedBy to /models
  • Added createdBy, updatedBy to /models/:id/dataBundles
  • Added createdTime, createdBy, updatedTime, updatedBy to /secrets
  • Added createdTime, createdBy, updatedTime, updatedBy to /transitions
  • Added createdTime, createdBy, updatedTime, updatedBy to /workflows
  • Added createdBy, updatedBy to /datasets
  • Added updatedTime, updatedBy to /organizations
  • Added retentionInDays to /documents. For documents with a datasetId, the minimum retentionInDays of the dataset and the document is chosen

2021-07-08​

  • Extended the period in which temporary credentials are valid upon first time invitation to Typenode or Flyt from 7 days to 30 days
  • Added timeoutInSeconds to /transitions
  • Fixed a bug preventing login to complete successfully in Typenode and Flyt
  • Fixed a bug preventing expired users from getting new temporary credentials when invited again

2021-07-01​

  • Added datasetId query parameter to DELETE /documents
  • Added datasetId query parameter to GET /documents
  • Fixed a bug preventing completion config in /workflows to be executed in some situations
  • Added datasetId to PATCH /documents/:id
  • Fixed error in JSON schema for /datasets, numDocuments -> numberOfDocuments

2021-06-25​

  • Deprecated /batches endpoint. It's replaced by /datasets. Your current batches will be unaffected until 2021-09-06, after which we will remove the entire endpoint and all of its data. Documents in batches will not be affected, only the batches themselves. Until 2021-09-06 you will not be able to create new batches. Please consider replacing your batches with datasets.
  • Added POST /datasets
  • Added GET /datasets
  • Added PATCH /datasets/:id
  • Added DELETE /datasets/:id
  • Added POST /models/:id/dataBundles
  • Added GET /models/:id/dataBundles
  • Added PATCH /models/:id/dataBundles/:id
  • Added DELETE /models/:id/dataBundles/:id
  • Fixed a bug causing incorrect error messages to be returned from the API
  • POST /transitions will now attempt to return appropriate status 400 error message when imageUrl is incomplete
  • PATCH /transitions/:id/executions/:id will now return status 400 error message when attempting to PATCH a timed out transition execution

2021-06-16​

  • Added loginUrls, defaultLoginUrl to PATCH /appClients/:id

2021-06-10​

  • Added GET /organizations/:id
  • Added PATCH /organizations/:id
  • Fixed a bug preventing the first log message to be written for workflow executions
  • Updated the email invite and verification code layout
  • Added numberOfRunningExecutions to /workflows

2021-05-26​

  • Added DELETE /models/:id
  • description in /models fieldConfig is no longer required

2021-05-19​

  • Added loginUrls, defaultLoginUrl to /appClients
  • Link to login button in invitation email now sends you to the app corresponding to the defaultLoginUrl in /appClients

2021-05-12​

  • Added PATCH /batches/:id
  • Added PATCH /appClients/:id
  • Added GET /models/:id
  • Added PATCH /models/:id
  • Added POST /models
  • Updated OAuth2 scopes
  • Added new possible value 'inactive' for status in /models

2021-04-27​

  • Added storageLocation, retentionInDays, containsPersonallyIdentifiableInformation to /batches
  • Added DELETE /batches/:id. Documents in batch must be deleted before deleting the batch
  • Added batchId query parameter to DELETE /documents

2021-04-23​

  • Added GET /logs. Use query parameters workflowId, workflowExecutionId, transitionId, transitionExecutionId to filter.
  • Added default retry configuration for workflow transitions that don't explicitly define one.
  • Added startTime, transitionExecutionId, workflowExecutionId to /logs

2021-04-14​

  • Now possible to create public app clients by using the generateSecret parameter set to false (defaults to true) and providing callback and logout urls
  • Added createdTime, apiKey, callbackUrls, logoutUrls, hasSecret to /appClients
  • Now preventing users from deleting themselves using DELETE /users/:id
  • Now preventing app clients from deleting itself using DELETE /appClients/:id

2021-04-13​

  • Added GET /batches
  • Added createdTime and numDocuments to /batches
  • Added POST /appClients
  • Added GET /appClients
  • Added DELETE /appClients/:id
  • Added createdTime, updatedTime, fieldConfig, preprocessConfig and status to /models

2021-04-09​

  • Added DELETE /secrets/:id
  • Added DELETE /assets/:id
  • Added paging to DELETE /documents. Supports deleting up to 1000 documents each API call.