Partilhar via


Cradl AI (Preview)

Automate internal document workflows with no-code AI.

This connector is available in the following products and regions:

Service Class Regions
Logic Apps Standard All Logic Apps regions except the following:
     -   Azure Government regions
     -   Azure China regions
     -   US Department of Defense (DoD)
Power Automate Premium All Power Automate regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Power Apps Premium All Power Apps regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Contact
Name Cradl AI Support
URL https://www.cradl.ai
Email support@cradl.ai
Connector Metadata
Publisher Ståle Zerener Haugnæss
Website https://www.cradl.ai
Privacy Policy https://docs.cradl.ai/legal/privacy-policy
Categories AI;Content and Files

Cradl AI makes it easy to automate internal document workflows with no-code AI. Extract data from any document automatically and free your organization from manual data entry.

Prerequisites

A free Cradl AI account. If you don't already have one, you can sign up here.

How to get credentials

See step 2 in Get started with your connector.

Get started with your connector

This quick start guide aims to provide a high-level overview of how Cradl AI can be used in a Power Automate workflow. Please refer to the official documentation for more details.

1. Creating a flow

Cradl AI Flows enables makes it easy to automate internal document processes in a simple, effective and unified way. Cradl AI's human-in-the-loop funcationality greatly reduces operational risk by automatically routing uncertain predictions to manual review. By adjusting your flow's confidence thresholds, you can decide when a document should be sent to manual verification and when it's allowed to pass straight through.

To create a new flow, go to Flows -> New flow.

2. Setting up a connection

In the Input section, click Add source and select Power Automate from the dropdown menu. Copy the text from the Credentials field and paste it into the Client Credentials field in your Cradl AI Connection.

3. Configure a webhook trigger

The easiest way to retreive the processed documents in Power Automate is by setting up a webhook trigger. Open your Cradl AI Flow and, in the Export section, select Webhook from the list of available integrations. Copy the HTTP POST URL from your webhook trigger in Power Automate and paste it into the Webhook integration in Cradl AI.

Known issues and limitations

See API Limits. Limits can be increased on request.

FAQ

Which document formats are supported?

JPEG, PNG, PDF and TIFF.

Can I use a single model for many layouts and document types?

Yes. In general, you need one model per document process. For example, if you want to automate an expense approval process where you process receipts, invoices and airline tickets, we recommend using one model even if you process multiple document types with different layouts.

How many documents do I need to train my model?

You will need at least 15 documents to train a custom AI model, but more training data is recommended for most use cases. Note that you do not need 15 examples of each layout. In fact, each document in your training dataset can potentially have a different layout.

Cradl AI's human-in-the-loop-functionality is a great way to improve your model over time by taking advantage of corrections from your business users to increase your training dataset.

Where is my data stored?

Our servers are located in Ireland in the European Union. Please refer to our to our Data Processing Agreement and Privacy Policy for more information about how personal data is processed.

Deployment Instructions

Refer the documentation here to deploy this connector as a custom connector in Microsoft Power Automate and Power Apps.

Creating a connection

The connector supports the following authentication types:

Default Parameters for creating connection. All regions Not shareable

Default

Applicable: All regions

Parameters for creating connection.

This is not shareable connection. If the power app is shared with another user, another user will be prompted to create new connection explicitly.

Name Type Description Required
Client Credentials securestring Your client credentials. True

Throttling Limits

Name Calls Renewal Period
API calls per connection 100 60 seconds

Actions

Create document

Extract data from documents like invoices, receipts, order confirmations, etc.

Parse document

Extract data from documents like invoices, receipts, order confirmations, etc.

Parse document with human-in-the-loop

Extract data from documents like invoices, receipts, order confirmations, etc.

Create document

Extract data from documents like invoices, receipts, order confirmations, etc.

Parameters

Name Key Required Type Description
Name
Name True string

The name of your document, for example Invoice.pdf.

File content.
File content. True binary

The file content.

Returns

Name Path Type Description
documentId
documentId string

The document ID.

Parse document

Extract data from documents like invoices, receipts, order confirmations, etc.

Parameters

Name Key Required Type Description
Model
modelId True string

Choose your model

Document ID
documentId True string

Name of the file, e.g. Document.pdf.

Output format.
outputFormat string

Output format.

The strategy used for aggregating predictions.
strategy string

Prediction aggregation strategy.

autoRotate
autoRotate boolean

A flag indicating whether document orientation should be auto detected.

maxPages
maxPages integer

The maximum number of pages to process in a single request.

imageQuality
imageQuality string

The image quality used when processing this document.

Returns

Name Path Type Description
documentId
documentId string

The document which was parsed.

inferenceTime
inferenceTime float

The number of seconds used to process this document.

modelId
modelId string

The model which was used.

Postprocessing.
postprocessConfig PostprocessConfig
Preprocessing.
preprocessConfig PreprocessConfig
predictions
predictions object

predictions

trainingId
trainingId string

The model training.

Parse document with human-in-the-loop

Extract data from documents like invoices, receipts, order confirmations, etc.

Parameters

Name Key Required Type Description
Workflow
WorkflowId True string

Choose your workflow

Document ID
documentId True string

The ID of the document which was submitted.

Title
title string

Title, e.g. Invoice.pdf.

Predictions
predictions object

Manually override the predictions used in this workflow execution.

Returns

Name Path Type Description
executionId
executionId string

A unique ID for this execution.

status
status string

A status indicator for this execution.

Definitions

PostprocessConfig

Name Path Type Description
Output format.
outputFormat string

Output format.

The strategy used for aggregating predictions.
strategy string

Prediction aggregation strategy.

PreprocessConfig

Name Path Type Description
autoRotate
autoRotate boolean

A flag indicating whether document orientation should be auto detected.

maxPages
maxPages integer

The maximum number of pages to process in a single request.

imageQuality
imageQuality string

The image quality used when processing this document.