Input and output data
Input
There are two options for how the Kayla AI Agent can ingest data:
From a shared file location
Via API
You can set the input method in the Input step during the AI Agent configuration.
If you choose the File ingestion input method, Kayla works with CSV files stored in the configured S3 location. The files must correspond to the following requirements:
They must contain a header row.
Any subsequent rows should each represent an entity to be reviewed, for instance, a pension plan or mutual fund.
They must include the Unique Identifier column containing the IDs of the entities for Kayla to review.
The column headers in the CSV files must match the field names you configure in the Input step.
If the input file is invalid, an error is thrown, and you can review it in Control Tower.
Output
As part of the default workflow, Kayla creates the following reports:
Entity level report created per entity (a row in an input file) after Kayla completes the review and human-in-the-loop tasks are closed.
Click the link to download a sample entity-level report.
API audit trail report generated only for the requirements that were completed using the data from API calls and did not include a supporting document from the source itself.
Click the link to download a sample audit trail report.
Batch summary report created per input file request (batch of reviewed entities). Once the batch is processed, it is emailed to the configured email address in the associated variation. The batch summary report is sent only if the Human-in-the-loop step is disabled in all cases.
Click the link to download a sample batch summary report.
Audit trail report for connector configurations generated by Kayla automatically for out-of-the-box connectors. For custom connectors, you can choose to enable or disable the report generation by setting the Audit trail generation parameter as you configure data providers.