AgentHack submission type
Agentic UI Automation
Name
Rafał Kozłowski
Team name
wavestack
Team members
How many agents do you use
One agent
Industry category in which use case would best fit in (Select up to 2 industries)
Logistic
Manufacturing
Complexity level
Intermediate
Summary (abstract)
We built an automation in UiPath that validates incoming invoices by matching them with existing purchase orders. The bot extracts data from the invoice, compares it with PO details like amounts and currency, then highlights any discrepancies. The result is a faster, more reliable invoice verification process with a clear status overview for the end user.
Detailed problem statement
Manual invoice verification is time-consuming, error-prone, and often creates bottlenecks in financial operations. When a company receives an invoice, someone must manually compare it to the related purchase order (PO) to verify details such as currency, tax amount, net amount and total amount. This repetitive task can lead to mistakes, delays in payments, and compliance issues. In large organizations handling dozens or hundreds of invoices daily, this manual process becomes a major inefficiency. Our automation aims to solve this by automatically reading invoice data, matching it against PO records, and instantly flagging any inconsistencies – allowing finance teams to act faster and with confidence.
Detailed solution
Our solution consists of a UiPath-based automation combined with a simple user interface that handles the entire invoice verification workflow. First, when a new invoice is received (PDF or scanned document), the robot extracts key data such as currency, tax amount, net amount, total amount, billing and vendor information using UiPath Document Understanding. This data is then saved to a central database. A second automation process checks each new invoice entry by looking up the corresponding purchase order based on PO number. It compares all critical fields to detect any mismatches. In case of discrepancies, the system flags the invoice and marks the validation status accordingly. The user interface allows finance staff to quickly see which invoices match POs and which require attention, with a detailed breakdown of what doesn’t align. This end-to-end process minimizes manual work, increases accuracy, and improves visibility into invoice processing.
Demo Video
Expected impact of this automation
This automation significantly reduces the time and effort required for invoice verification. On average, manual comparison of an invoice to a purchase order takes 5–10 minutes per document. Our solution reduces this to under 1 minute, saving hours of repetitive work each week. In a company processing 100 invoices per day, this can translate to over 30 hours of work saved weekly. The automation also minimizes human error, improving compliance with procurement policies and reducing the risk of incorrect payments. Additionally, finance teams gain better visibility and control over the invoice approval process, leading to faster processing times and stronger vendor relationships. The expected ROI includes both labor cost savings and operational efficiency gains, especially for organizations with high invoice volumes.
UiPath products used (select up to 4 items)
UiPath Apps
UiPath Data Service
UiPath Document Understanding™
UiPath Studio
Integration with external technologies
Data Services from UiPath
Agentic solution architecture (file size up to 4 MB)
Sample inputs and outputs for solution execution
The process starts when a user opens the application and uploads an invoice file (PDF or scanned image). The system automatically extracts key data from the invoice, such as the PO number, currency, net amount, tax, and total amount. It then searches the internal database for the corresponding purchase order and compares all relevant fields. Based on this comparison, the application assigns a status — either “Correct” or “Need attention”. If there are any discrepancies, the user can open a detailed view that clearly shows which fields don’t align with the PO (e.g., net amount or currency). This allows finance staff to quickly validate the invoice or take corrective action when needed.