OCR Services

Purpose

The OCR Services page in the Admin Console sets the optical character recognition (OCR) provider that DocCenter, advanced IDP tools AI skill, and document-based generative AI skills use across your environment. See What the OCR provider setting controls for a full list.

By default, Appian uses its own infrastructure for OCR. If you have an Amazon Textract, Azure AI Document Intelligence, or Google Cloud Document AI account, you can set one of them as your OCR provider.

Access

To access the OCR Services page, go to the Admin Console. Then, in the left navigation menu, under Artificial Intelligence, click OCR Services.

How the OCR provider affects AI skills

The OCR provider affects different AI skills in different ways. Review the following before configuring a provider.

What the OCR provider setting controls

When you set an active provider, all advanced IDP tools AI skills and document-based generative AI skills in your environment route through that provider. DocCenter extraction models that use Bounding Boxes also use the active OCR provider.

The OCR provider setting is independent of the AI model provider setting. Changing the AI model provider doesn't change where OCR runs, and changing the OCR provider doesn't affect which large language models your skills use. For details on which skills are affected by each setting, see How the providers apply to AI features.

The active OCR provider applies to the following AI skills:

• The advanced IDP tools AI skill (see Using advanced IDP tools with non-Appian providers for important information about limitations)
• The following document-based generative AI skills:
  • Unstructured document extraction (Text only mode)
  • Document redaction
  • Document text summarization (Text only mode)
  • Document text PII detection (Text only mode)

The OCR provider doesn't affect the following AI skills:

• Prompt builder.
• Classification.
• Semi or highly structured extraction.
• Spreadsheet extraction.
• Generation.
• Text and email skills for classification, extraction, summarization, and PII detection.
• Visual elements & text mode for unstructured document extraction, document text summarization, and document text PII detection.

Using advanced IDP tools with non-Appian providers

When Microsoft Azure or Google Cloud Platform is the active provider, the advanced IDP tools AI skill supports only the Text tool. All tools listed under "Usage limits apply" (Expense Data, Key-Value Pairs, Checkboxes, Tables, Signatures, and Search Queries) cause the test or smart service execution to fail.

To resolve the failure, open the AI skill, turn off the unsupported tools, and save. The process model outputs for deselected tools return null, so update any downstream process nodes that depend on those outputs.

Language support

The languages your AI skills can process for OCR depend on the active provider. If Appian or Amazon Web Services is your active provider, they both use Amazon Textract and support the same languages. For the full list, see Supported OCR languages. If you use a different provider, check the provider's documentation for their supported languages.

Available providers

The OCR Services page displays the following providers:

The active provider displays an Active badge on its card.

Configure a provider

Select a provider card to configure its credentials.

Amazon Web Services

Choosing Amazon Web Services as your active OCR provider uses Amazon Textract for text recognition. Appian's default OCR also uses Textract, so switching to Amazon Web Services gives you direct control over the Textract account without changing OCR behavior.

To configure Amazon Web Services as an OCR provider:

1. On the OCR Services page, select Amazon Web Services.
2. Enter the following credentials:
   • Access ID: Your AWS access key identifier.
   • Secret Access Key: Your AWS secret access key.
   • Region: The AWS region where your Textract resources are located.
3. Click TEST CONNECTION to validate your credentials.
4. Click SAVE.

Amazon Web Services IAM permissions

Before configuring Amazon Web Services as an OCR provider, your AWS administrator must create an IAM user with the following permissions:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Action": [
                "textract:DetectDocumentText",
                "textract:AnalyzeDocument",
                "textract:AnalyzeExpense"
            ],
            "Effect": "Allow",
            "Resource": [
                "*"
            ],
            "Sid": "TextractPermissions"
        }
    ]
}

Microsoft Azure

Choosing Microsoft Azure as your active OCR provider uses Azure AI Document Intelligence for text recognition.

To configure Microsoft Azure as an OCR provider:

1. On the OCR Services page, select Microsoft Azure.
2. Enter the following credentials:
   • Endpoint: The base URL for your Azure AI Document Intelligence resource.
   • API Key: The access key for your Azure resource.
3. Click TEST CONNECTION to validate your credentials.
4. Click SAVE.

Google Cloud Platform

Choosing Google Cloud Platform as your active OCR provider uses Google Cloud Document AI for text recognition.

To configure Google Cloud Platform as an OCR provider:

1. On the OCR Services page, select Google Cloud Platform.
2. Enter the following credentials:
   • Project ID: Your Google Cloud project identifier.
   • Location: The Google Cloud region for your Document AI processor.
   • OCR Processor ID: The identifier of your Document AI processor.
   • Service Account Key: The service account key for authenticating to your Google Cloud project.
3. Click TEST CONNECTION to validate your credentials.
4. Click SAVE.

Set the active OCR provider

After saving a provider's credentials, you can set it as the active OCR provider for your environment.

To set the active OCR provider:

1. On the OCR Services page, select a provider with saved credentials.
2. Click SET AS OCR PROVIDER.
3. Review the confirmation, then click SET AS PROVIDER.

The provider card displays an Active badge to indicate it's the current OCR provider.

Clear saved credentials

You can't clear the credentials of the active OCR provider. To clear an active provider's credentials, set a different provider as active first.

To remove a non-Appian provider's saved credentials:

1. On the OCR Services page, select the provider.
2. Click CLEAR CREDENTIALS.
3. Review the confirmation, then click CLEAR CREDENTIALS.

Feedback