SI-Bone serves hospitals and surgeons, with field sales staff promoting the use of SI-Bone's implants during surgeries. Following surgery, the sales team compiles a request for a purchase order (RPO), submitted to the hospital, leading to the issuance of a purchase order (PO). These POs arrive in various formats (hardcopy attached to email, handwritten RPO with handwritten PO, handwritten RPO with typed PO, and email body) to SI-Bone’s finance team who are tasked with invoice creation. The team manually reviews each email and attachment, identifies the PO and RPO, checks the ERP for corresponding sales orders, issues invoices and updates the ERP with relevant PO details including attaching them.
The entire process was time-consuming, taking 5-30 minutes to process a purchase order and issue an invoice. This delay was mainly attributed to the manual task of opening each email and attachment, along with the need for manual validation against the ERP. These manual steps introduced the risk of errors, resulting in rework and invoice reissuing. Given the high volume of purchase orders during month or quarter close, the team had to work extended hours to handle the increased workload.
Recognizing the potential for automation, Accellor proposed the implementation of an AI-based intelligent automation solution to streamline the entire process of PO submission and invoice generation, aiming to reduce human interaction. This solution automated the identification of crucial details (including RPO number, PO number, surgeon and hospital names, and date of surgery) from incoming emails. It validated this information against the ERP, issued invoices, and updated the ERP with invoice and PO details, including attachments. The implementation included training models to identify relevant fields in both emails and attachments, along with programming the automation workflow. Additionally, we ensured that the new solution complied with regulatory requirements and was SOX compliant. While the solution handled the majority of POs, human intervention was necessary for exceptional cases, such as illegible handwriting or typos leading to the interchange of PO/RPO details in emails.
The intelligent automation solution implemented for processing purchase orders (PO) successfully reduced processing times from 5-30 minutes to a mere 30 seconds, resulting in significant time and cost savings. Moreover, the automated process requires no human oversight unless there is an exception. The automated processes also minimized the potential for manual errors, allowing the team to swiftly handle all incoming purchase orders without the necessity for extended working hours during month/quarter close.
