Set up a use case for Now Assist in Document Intelligence
Create a use case record to define a document you want to process with Now Assist in Document Intelligence.
始める前に
- Activate a Now Assist in Document Intelligence skill. For more information, see Activate a Now Assist in Document Intelligence skill.
- Role required: Admin, DocIntel Admin, DocIntel Manager
このタスクについて
In a use case, you define the information you want Now Assist to get from a document by specifying the type of document to process, the fields and tables to detect or questions you want to ask, and the location where document processing results are stored.
Once you have defined a use case, users can begin processing documents for it in the related workflows.
For more information on use cases, see Use cases in Now Assist in Document Intelligence.
手順
- Navigate to All > Now Assist Admin > Skills.
- In the workflow list, select Platform.
-
In the Platform skills list, find the applicable document intelligence skill and select Edit in the options menu (
).
-
Select New use case.
The guided use case setup opens.
-
Define the use case.
- Enter a name for the use case.
- Select a target table to store the document processing results for this use case.
-
Select the language of the files processed for the use case.
If the files contain multiple languages, select the primary language.
For more information, see Languages supported by Now Assist in Document Intelligence.
-
Select the large language model (LLM) that will make predictions for the documents processed with this use case.
For more information, see Large language models used by Now Assist in Document Intelligence.
- オプション:
Turn on image mode to process images more efficiently.
Image mode sends pages to the LLM as images to leverage the visual capability of the multimodal LLM as well as any of the languages supported by it.
The image mode option is available when a multimodal LLM is selected.
注:Selecting image mode reduces the page count limit to 10 pages per file. - Select Save and continue.
-
Define information or questions.
-
Select Add a field.
If you have already defined one or more fields, tables, or questions and you want to add another, select New field.
-
Select the type of information you want to get from the document.
You can choose one of the following:
- Field
- Fields are used to extract a single piece of information in the document. For example, a document number or a customer name.
- Table
- Tables are used to extract lists or tables of information. A table can have multiple columns. The number of list items or table rows doesn’t have to be known in advance.
- Question
- Define the question you want to ask about the document.
A form displays based on the information type you selected. -
On the form, fill in the fields.
The type of form depends on the type of field.
図 : 1. Field form for document extraction -
Select Save.
The system adds the new fields, tables, or questions to the Information list associated with the use case.
- オプション:
To edit or delete a field, table, or question, select the appropriate option in the options menu (
) of the row.
- Select Save and Continue.
-
Select Add a field.
-
Test the use case performance with a sample document.
Select Test a new document.
-
Select a document.
Option Description Upload from record - Select Upload from record.
- Enter search criteria in the search field.
- Select a record from the list.
This option is available when a target table is selected for the use case.
Upload from this device - Select Upload from this device.
- Select Add file.
- Select a file and select Open.
- Select Upload.
-
Select Continue.
ヒント:Select the Open in a new tab button (
) to view the document in a larger workspace on a separate browser tab.
The Document Intelligence workspace appears in a frame on the Test output screen. - Review the performance of the skill for the test document.
- Select Save and continue.
-
Select a document.
-
Add integrations.
Integrate the document intelligence use case with a workflow.
This option is available when a target table is selected for the use case.
For more information on Document Intelligence integrations, see Integrate with a custom application or workflow.
-
Select Add integration.
If you have already defined one or more integrations and you want to add another, select New integration.
- Enter a name for the integration.
-
Select the type of integration you want to use.
The Process task type creates an integration point to automatically create and process document tasks based on specific triggers happening in the target table.
The Extract values type creates an integration point to automatically propagate the extracted values to the target table when extraction has been completed in Now Assist in Document Intelligence.
-
Use the conditions to select certain fields as specific triggers for the integration.
Conditions are available if you selected Process task in the previous step. For more information on conditions, see OR conditions.
- オプション:
Select the Create Flow option to create a flow for this integration in Workflow Studio.
ヒント:This option should be selected, unless you are planning to write your own custom script to set up the integration. Be sure the integration is activated on Workflow Studio. For more information, see Building flows.
- Select Save.
- Select Save and continue.
-
Select Add integration.
-
Review and activate.
- Review the configuration selections you’ve made for the use case.
- オプション: Select Back to return to a previous step and make a change
- Select Complete setup.