Traditionally the Procure to Pay (P2P) process begins with the requisition of goods and services and ends with the payment being issued to vendors as a part of the accounts payable process. Accounts Payable (AP) is a subprocess within the wider spectrum of the P2P which focuses on Invoice handling, by an Accounts Payable (AP) persona, before approving the invoice for payment processing.
The manual steps followed by the AP persona to approve an invoice are -
- Captures invoices as soon as they are received via email or fax.
- Reads the invoice and imports the data manually into an IT system.
- Retrieves the associated purchase order and or receipt.
- Confirms that all the charges are accurate.
- Connects with the ERP system to issue or release the payment to the vendor.
The manual process, as defined above, comes with its set of challenges: Poor or erroneous data entry, Inefficient storage and messy record keeping, Slow invoice processing, Inefficient fraud detection, duplicate payments, Lack of end-to-end visibility of the entire invoice handling process.
We have automated the above process using various capabilities from the Cloud Pak for Business Automation platform and AI/ML micro-services with a goal to achieve touch-less invoice processing.
The automated end to end solution includes the following key capabilities:
- Invoice content digitization using IBM DataCap and Intelligent Extractor(a Datacap asset built by IBM Expert Labs) – We have trained huge Invoice Document sets from various vendors with 3 to 4 invoice variations per vendor. We have attained an extraction accuracy of 77%
- Automated decisions driven by business rules including the 2-way and 3-way invoice matching rules using ODM.
- AI/ML micro-services to enable straight through processing. The Invoice goes through the various AI/ML enabled features - vendor identification, Advanced Risk Management to identify duplicate invoices, semantic match of Invoice and PO line, GL Coding, and Tax code prediction for line items using AI micro-services
- Orchestration of the AI micro-services using WDG
- Human in the loop for exception handling and providing feedback to the machine learning (ML) models for continuous improvement using BAW
- Business Insights (KPIs) using BAI
The End-to-End solution view can be seen below –
The entire solution is modular, customers can choose to include individual blocks in the diagram independently to start their automation journey.
By deploying this automated solution, customers will see significant improvement with the AP processor’s productivity and invoice processing cost each year
We currently have the Invoice digitization and the AI/ML micro-services part of the solution in production. We are seeing an extraction accuracy of 77% and around 75% of the invoice going touch-less.
#BusinessAutomationInsights(BAI)#BusinessAutomationWorkflow(BAW)#FileNet#ibmautomation#ibmcloudpakforbusinessautomation#IBMDatacap#industryusecase#OperationalDecisionManager(ODM)#UseCase#WDG