Oracle Purchase Order Cleanup with UiPath

Automating the review and cleanup of 30,000 purchase order records through an on-premises Oracle user interface.

Business challenge

A business needed to review and clean up approximately 30,000 purchase order records in its on-premises Oracle system.

Each record had to be opened individually, reviewed against specific values and assessed for the appropriate cleanup action. Navigation and verification took approximately three to six minutes per record, with three users expected to work through the backlog.

A further 800–1,500 incorrect entries arose each month, creating an ongoing requirement for repeatable cleanup.

Use case objective

Automate purchase order validation and cleanup using pre-approved conditions and value parameters, while referring exceptions for human review.

The solution needed to operate through the existing Oracle frontend, resolve the initial backlog and remain available to trigger when further cleanup was required.

Process before automation

1. A user logged into the on-premises Oracle system.
2. They navigated to and opened an individual purchase order record.
3. They reviewed the relevant values against the cleanup criteria.
4. They determined and performed the appropriate cleanup action.
5. They repeated the process for each remaining record.

At approximately 30,000 entries, this represented a substantial workload of repetitive navigation, verification and data correction.

Automation scope

Kyberus implemented a UiPath frontend automation that logged into the on-premises Oracle environment and navigated the application through its existing screens.

The bot applied pre-approved conditional rules and value parameters to validate each record and determine the permitted cleanup action.

Built initially for a once-off backlog cleanup, the automation remained available for subsequent on-demand runs.

Automated workflow

1. Trigger: on-demand cleanup

The automation was initiated when a purchase order cleanup was required.

2. On-premises login and navigation

UiPath logged into the Oracle environment and navigated to the relevant purchase order records through the application’s user interface.

3. Record validation

The bot opened each record and checked its values against the pre-approved conditions.

Control point: Cleanup decisions were restricted to the agreed validation rules and value parameters.

4. Frontend cleanup

For records that met the approved criteria, the bot performed the relevant cleanup actions through the Oracle screens.

5. Exception referral

Records that could not be resolved within the approved conditions were flagged for manual review.

Business outcomes

Initial cleanup run completed in three days: The automation ran continuously for approximately 72 hours to work through the 30,000-record scope.

200 exceptions identified: Approximately 0.67% of records were flagged for human review, concentrating manual attention on a small subset of the original workload.

Approximately 99.3% processed without an exception: The remaining records passed through the automation without being flagged for manual assessment.

Reusable cleanup capability: The workflow could be triggered again to address subsequent incorrect entries.

Existing application workflow automated: UiPath handled login, navigation, verification and cleanup through the on-premises Oracle frontend.

Time saved

Manually reviewing and cleaning up 30,000 Oracle purchase order records would have taken an estimated 1,500–3,000 staff hours, based on three to six minutes per entry.

For three users dedicated to the task, that equates to approximately 63–125 working days. The UiPath frontend automation completed the processing run in three calendar days, flagging just 200 exceptions for manual review.

Allowing an estimated 10–20 hours to review those exceptions, the automation potentially saved 1,490–2,980 hours of manual processing effort.

For the additional 800–1,500 incorrect entries arising each month, running the automation again could address a further 40–150 hours of monthly manual workload.

Estimates exclude automation development, supervision, final verification and additional exception remediation. The three-user comparison assumes eight-hour working days. Machine runtime is separate from staff effort.

Use case applicability

This approach is relevant to businesses that:

• Manage purchase orders in an on-premises Oracle environment.
• Have large backlogs requiring record-by-record review.
• Perform repetitive checks and corrections through application screens.
• Can define clear, approved cleanup conditions.
• Need human review for exceptions.
• Require a repeatable, on-demand cleanup process.

Explore Oracle automation with UiPath

Kyberus helps businesses automate repetitive procurement and administrative tasks using UiPath. If your team is manually navigating Oracle to review and correct thousands of records, we can assess which steps can be automated and where human judgement remains necessary.