Now Assist in Doc Intel - Not able to use doc intel output in the now assist custom skill

[MDBilalI]

Hello All,

I have created a doc intel use case and in that created some fields, perform some tests by uploading some documents and it is extracting the values correctly.
Then I added that doc intel use case in a now assist custom skill as a tool, but skill is not able fetch or use the doc intel output.
If you have any idea about that, please provide your response.

Thank you, 

Bilal

[KPNow]

Hi  @MDBilalI 

This issue could stem from a timing gap and output format mismatch: Document Intelligence processes extractions asynchronously, while Now Assist skills expect immediate, structured data. If your custom tool triggers Doc Intel and immediately attempts to read the response—or simply passes the document record sys_id without querying the extracted results—the skill receives an empty payload or an "in-progress" status before the extraction pipeline finishes. Furthermore, Now Assist cannot automatically read raw Doc Intel task records unless the tool explicitly parses the key-value pairs out of the underlying extraction tables (such as sys_docintel_value or via the sn_docintel scriptable API) into defined output variables.

To fix this, back your custom tool with a custom Script Action or Flow Action that acts as a structured middleware. Have the action accept the target document/record ID, query for the completed Doc Intel extraction task, and construct a clean JSON object containing only the extracted key-value fields. Ensure the Flow Action waits for the extraction status to hit Completed before returning, or ensure you only invoke the Now Assist skill after the document status updates to processed. Finally, verify that your Now Assist skill tool output parameters explicitly map to these JSON keys, and that the GenAI/system execution context has appropriate ACL read permissions on the Doc Intel extraction tables. 
Hope this helps you.