With the recent release of IBM RPA and IBM Automation Document Processing (ADP), we enabled new possibilities on helping our customers to automate their document processing workflow. In this 2 part series, we are going to explore some of the new features in the offerings, and also why we want to use them together.
Imagine a typical order processing workflow where we have a human operators responsible for entering purchase order received into the Sales Order system (e.g. SAP) manually. To automate that procedure, there can be different routes depending on how the purchase order are received.
One option is for the company can rebuild the entire workflow by creating an end-to-end integration by first providing a website or a mobile application for their customers to enter the order information. Once the order is received, it can trigger an integration flow to call the backend SAP services via API. This will work if company is willing to invest in digitizing their end-to-end customer journey and if they can change how other customers will create purchase orders. However, for many situations, it might be difficult to justify the cost of such re-engineering effort or what if there are real business needs to receive information through non-digitalized means - e.g. fax, scanned photos, etc? That is where one can use a combination of RPA and document process technology to solve.
Traditional document processing technology is rules-driven - i.e. content extraction is programmed ahead of time based on a set of pre-defined extraction rules. This approach works well if the document format are largely the same. On the other hand, the reality of content like Purchase Orders are largely depend on the format of the purchasers and not the suppliers. This means that for rules-based content extraction to be successful, the system must be pre-programmed to recognized all the formats - and if certain formats are not recognizable, the workflow will still have to delegate the processing back to human operators to perform the order entry.
With IBM Automation Document Processing, our goal is to combine rules-based content extraction with deep-learning AI to read and extract information from documents.
Using the low-code supervised training technique as part of the workflow, our goal is that over time, we will be able to increase our extraction accuracy and content variations without having to go back to re-program the extraction rules.
In the next part, Mira Kim, Offering Manager for Document Processing will talk about
how ADP can be used to help automate utility bill payment processing.