Skip to main content
Skip table of contents

Retrieve Payable (AI)

Part of the Transform AI stack, the retrieve node collects processed document data from Square 9’s AI Cloud.

Minimum configuration for the node are limited to providing a Release Channel and a license key. Unless otherwise specified, all customers should be running production workflows on the channel named “stable”. A “labs” channel is also available for test/preview features, but Square 9 makes no guarantees about labs not introducing breaking changes to a workflow process.

In scenarios where extracting invoice line items is desirable, a Table Field can be specified for line items and extracted line item fields can be mapped to fields of the identified table. Use the “Clear table before update” option in scenarios where the table might get reused in a process and you need to ensure it’s empty when the Node executes at this step. If you don’t clear it, data is appended to anything already existing in the table.

Data cleanup options are available in the node, documented here.

Making changes to clean up options for published workflows can impact future results. It is suggested changes here are thoroughly vetted.

The Retrieve node does not trigger any conversations with Square 9’s AI Cloud after it’s first request for a process. In other words, if the node has been encountered once on a process, the data retrieved will be collected and cached for reuse in the workflow for that process. If you need to run AI extraction again in the process, you must first use the Post Payable (AI) node.

Version History





Initial Release



Adding production/stable support for line items.



Adding support for collecting ORIGINAL_VENDOR_NAME for use in retraining of machine learning for vendors. Dynamically mapping Header/Footer fields are no longer case sensitive.

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.