UiPath Robot Reconciles EnergyLink Partner Invoices for a US Producer

The client is a US oil and gas producer listed on the NYSE. It is growing fast, mostly by acquiring other companies, and during the project the group counted roughly 120 companies.

In the US oil and gas industry, operators invoice their partners with joint interest billing (JIB) invoices, and they report production and revenue per well. Our client received all of this through EnergyLink, an Enverus platform where operators publish invoices and revenue statements to their partners. Each month someone had to check the invoices against the well data the company kept itself, and each step was done by hand.

The reconciliation itself could not start until someone had logged into EnergyLink and pulled five reports per company, every report with different filters. Across roughly 120 companies, that alone took about ten working days each month.

The real cost

• Ten days a month spent downloading. Some 600 reports pulled before a single line was reconciled.
• One full-time role doing mechanical work. Downloads plus cross-checking added up to a whole FTE.
• Mismatches caught late. Gaps between invoices and well data only appeared once every download was done, and following them up across teams happened by hand.
• Each acquisition added to the load. New companies and new wells joined the monthly cycle with every deal.
• Scanned paper in the flow. Some invoice data came in as scans, which do not yield to simple extraction.

The monthly reconciliation now runs with nobody downloading reports. The team only sees discrepancies, and each one arrives already assigned.

• A fully automated monthly process replaced one full-time role of manual work.
• Roughly ten working days of downloads gone each month, about 120 working days over a year.
• Every month, around 120 companies and five reports for each are covered.
• Discrepancies land with the right team in Action Center, with no email chasing.
• Delivered in roughly a month. We estimate, from the value of the role it replaced, that the build pays back within a year. ROI was not measured by the client.

Where reconciliation robots fail

Coverage decides how good a reconciliation robot is, and in a company that grows by acquisition the entity list changes each quarter. Hard-code last year's companies and the robot quietly misses the new ones. Scanned invoices are the other trap: a small slice of volume that eats most of the effort. Keep entities and report filters in configuration, send anything unreadable or unmatched to a person, and no entity drops out of the monthly close as the group grows.

One more UiPath and Action Center project for a US company: processing shipping documents for an industrial distributor.