Product Information
Business document processing with SAP AI Business Services: consumption options
Credit to Robert Mancine and Tim Nusch who helped me drafting this blog post.
In this blog I briefly introduce the topic of business document processing and outline three ways how to consume the related SAP AI Business Services. Subsequently, I will post a short series of further blogs, in which I will explain each and every consumption approach in more detail and provide additional assets for your reference. The aim is to give you a full picture of what is possible with SAP AI Business Services.
Business Document Processing
Despite various ways of electronically exchanging information, enterprises still process large amounts of documents, such as supplier invoices or customer orders, manually. Often these documents are unstructured and vary from document to document – making it a challenge to automatically transform and embed the information directly into business processes. Business document processing aims to automate this process, generating instant business value. Figure 1 depicts typical steps when processing business documents in enterprises (see Fig. 1):
- Classification: understanding the type of a document to determine what to do with it.
- Extraction: extracting business relevant information out of a document to use it in the subsequent business process.
- Enrichment: enriching extracted information with additional master data and posting it back to the system of record for further processing.
Fig. 1: Business document processing
There are three SAP AI Business Services for business document processing: Document Classification, Document Information Extraction, and Business Entity Recognition. They are all available on SAP Business Technology Platform. A detailed description and the relevant use cases can be found in my previous blog post: Simplify Business Document Processing with SAP AI Business Services.
Here are the three common ways how to consume AI-enabled business document processing by SAP.
Standard Integration
Business users expect AI capabilities without the need to integrate, run data science projects or execute costly training activities for machine learning. This type of integration is referred to as ‘everyday AI’. This simplification can be achieved through embedding the technology directly into the business process where the data resides. In fact, recent research has shown that this approach can lead to higher user satisfaction, i.e., the fulfillment of user expectations. However, the main challenge remains finding the right purpose with perceivable business value (i.e. relevant business use case) as well as delivering the technology in a compliant way, as part of product or process innovations (Janasz et al., 2021).
In recent years SAP has identified several business relevant use cases related to business document processing and implemented them successfully in multiple business applications (e.g. SAP S/4HANA) by reusing standard SAP AI Business Services, such as Document Information Extraction. This means that SAP users can consume technology within the business process without even noticing it. The most popular busienss document processing use cases are related to:
- accounts payable,
- accounts receivable, and
- sales order creation.
In my first blog post in this series I consolidated all such scenarios. There you will find the list of relevant SAP products, the use cases, as well as links to additional assets such as: SAP Help Portal, demo videos, and further blog posts. You will find the link here and also at the end of this blog.
Low-Code/No-Code Automation
AI in conjunction with Robotic Process Automation (RPA) has also ushered in myriad possibilities for custom process automation. Enterprises that possess access to SAP Business Technology Platform can develop rapid and custom automation scenarios in low-code/no-code manner. The foundation for that is that SAP AI Business Services are fully compatible with SAP Intelligent RPA. You can connect to the APIs as part of the bot flow, allowing for end-to-end automation scenarios, such as automating the processing of email attachments or extracting information from PDFs.
In my second blog post I will discuss several common scenarios that already have accelerators available allowing customers to quickly implement the solutions and begin getting business value. I will also discuss the advancements in SAP Intelligent RPA that make it even easier to build automation for processing unstructured documents.
Full Integration Flexibility
Finally, business and IT departments of enterprises require the ability to extend their applications and further automate business processes by infusing business document processing capabilities into the existing landscape. Business processes can run accross multiple applications in complex hybrid architectures. Hence, enterprises require ‘everyday AI’ to be consumed in a compliant, secure, and hybrid way, but without replacing established applications.
In my last blog of this series I will present some ideas how to flexibly compose own integration scenarios on SAP Business Technology Platform by utilizing a range of customizable tools such as: SAP Workflow Management, Business Rules, or Cloud Integration. I will also share integration packages that are available for certain use cases on SAP API Business Hub.
What’s next?
Here is the list of blog posts related to this series:
- Business document processing through standard SAP integration.
- Business document processing through low-code/no-code automation.
- Business document processing and full integration flexibility.
These blogs will be updated as soon as new consumption options become available. For more information on SAP AI Business Services, please check out the below links.
Explore: SAP Community Page
Dive deeper: Open SAP Course
Get an overview: Blogpost part I | Blogpost part II
Understand Scenarios: Intelligent AI Use Cases
Exchange Knowledge:
Document Classification Questions | Document Information Extraction Questions
Business Entity Recognition Questions | Service Ticket Intelligence Questions
Data Attribute Recommendation Questions | Invoice Object Recommendation Questions
Source:
Janasz T., Mortensen P., Reisswig C., Weller T., Herrmann M., Crnoja I., Höhne J. (2021): Advancements in ML-Enabled Intelligent Document Processing and How to Overcome Adoption Challenges in Enterprises; In Swiss Journal of Business Research and Practice 75(3), 340-358.