Working With Exception Management in Sandbox Accounts

You can enable Exception Management (EM) in Sandbox to run inference using models trained in the corresponding Production account. This lets you safely evaluate EM workflows with production-quality results without duplicating model training or requiring Sandbox data to meet historical volume thresholds. No model training occurs in Sandbox. Standard EM workflows, navigation, and actions remain the same as in Production, for example, tabs for Transaction Errors and Expected Transactions, KPIs, and quick actions.

Important:

Exception Management is a limited release. Availability depends on data volume and model readiness. Models are trained in Production only and typically require approximately 18 months of historical transaction data.

In Sandbox, only Sandbox transactions are sent for inference to the Production-trained models: an initial two-week backfill runs on enablement, then hourly on newly created or edited transactions.

Standard EM permissions and role-based visibility apply, as in Production.

Prerequisites

  • Ensure Exception Management AI Models is enabled in your Production account. Administrator role is required to enable features.

  • Production account must have sufficient historical transaction data to train the models.

To use EM in Sandbox:

  1. Enable Exception Management AI Models in Production (Administrator)

    • Go to Setup > Company > Enable Features > Accounting.

    • Under Advanced Features, turn on Exception Management Models.

    • Click Save. If prompted, confirm that this is intended for SandBox testing.

  2. Enable Exception Management in your Sandbox

    • In the Sandbox account, go to Setup > Company > Enable Features > Accounting.

    • Under Advanced Features, turn on Exception Management to render the EM UI in Sandbox and connect to the Production-trained models.

    • Click Save.

  3. Access EM and review exceptions in Sandbox

    • Go to Lists > Accounting > Exception Management.

    • Use the two subtabs to review flagged exceptions:

      • Transaction Errors: Displays Incorrect Accounts, Incorrect Amounts, and Vendor Information Change, grouped by transaction and business flow.

      • Expected Transactions: Displays Missing Transactions.

    • On a selected item, click View Details to see why the item was flagged, view similar past transactions, and use quick actions to resolve the issue. For more details, see Managing Exceptions.

Note:

If EM options aren't visible in Sandbox, ensure Exception Management AI Models is enabled in the corresponding Production account, then reopen Enable Features in Sandbox. Additionally, if no exceptions appear in Sandbox, ensure there is recent Sandbox activity, wait for the hourly cycle, and check role permissions.

Related Topics:

General Notices