IBM has rolled out a new open source Software Developer Kit (SDK) that lets users call Watson services directly from ABAP code in SAP systems.
The project’s lead developer, IBM’s Bradley Knapp described the release to Computer Business Review as a “perfect entrance point for the democratisation of AI and machine learning for the SAP ERP space.”
New IBM ABAP SDK: Handy for Automating ERP Processes
ABAB, or “Advanced Business Application Programming” is a high-level programming language created by the German software company SAP.
It is the primary programming language supported on the SAP NetWeaver ABAP application server platform and applications that run on it, such as SAP ERP and S/4HANA. The new SDK lets developers tap IBM’s machine learning services on Watson using thousands of lines of pre-built ABAP classes, methods, data types; responses are translated back to ABAP.
IBM’s Bradley Knapp told Computer Business Review: “By allowing SAP developers to use their native language to interact with Watson Developer services, it’s now simple for them to bring all the benefits of AI into their existing processes and workflows to improve data integrity, automate previously (terrible) manual tasks, and most importantly, start their journey to actually using AI and ML to transform their business.”
The aim of the SDK is to strip out a layer of complex python, and let SAP developers interact with Watson services natively with the data models that they already have created and stored. An example?
Knapp said: “From a business perspective, imagine the following scenario: A US-based global company has to deal with invoices from all over the world. Many of them are delivered electronically and in English, and are super easy to programmatically enter into their ERP system.
“But some of those invoices might be handwritten, and in Spanish. In the old world, you’d have to have a manual intervention where a person would have to look at the scan of the invoice, translate it into English, and then manually enter the data into the SAP system.
“Now, the ABAP programmer can send that invoice scan to Watson Visual Recognition for Optical Character Recognition, and then the OCR data that’s returned can be sent into Watson Natural Language Processing for translation from Spanish to English for population into the ERP.
“The greatest benefit is that the speed of processing of this data is dramatically improved – from days to seconds, increasing responsiveness and data cleanliness and timeliness, which is always of primary concern for an ERP system… [and the person previously doing this] manual intervention is now freed from this manual task.”