Robotic Process Automation’s Role in Pandemic Recovery

RPA can also be combined with conversational AI technologies, which apply natural language to communicate intent and trigger actions downstream. Under the new rules of COVID-19, wherein businesses are challenged to interact physically, the need for digital customer engagement and service has been amplified.

RPA Cloud ERP Co-Innovation Project in China

SAP views RPA as an essential technology in its goal to help customers realize the power of the Intelligent Enterprise. With prescient foresight in 2019, Jan Gilg, president SAP S/4HANA, called on development teams to make bot development a priority. SAP employees based in China took on the challenge, and the fruits of their labor were recently rewarded when REHAU Polymer China went live in June with an RPA bot and machine learning. The application automated invoice processing and helped REHAU achieve higher process efficiencies.

REHAU is the first SAP S/4HANA Cloud customer in Greater China, as well as the first reference customer for SAP Intelligent RPA and intelligent technology packages that include machine learning, natural language understanding, and advanced analytics capabilities. The co-innovation project with REHAU brought together teams from across SAP: SAP S/4HANA Finance & Risk, New Ventures & Technologies, and Intelligent Delivery Group of Greater China.

To streamline and automate the financial operational process in SAP S/4HANA Cloud, project members trained a machine learning model that, when used in combination with the RPA bot, was able to achieve greater accuracy and seamless automation.

A machine learning model is only as good as the data it has been trained on and each method of training – isolated, centralized, or federated – has its own advantages and disadvantages. Certain types of data, such as transaction data, is highly sensitive and usually cannot be shared outside a company’s network for building a model, which makes it difficult to apply AI for RPA solutions.

“While training the model, we were able to keep sensitive data on the customer’s system, aggregating it in anonymous form to build a federation of data,” shares Voga Li, data scientist at SAP Labs China. “This method allowed us to achieve a higher prediction accuracy for the model.”

The REHAU project under the guidance of SAP demonstrated that federated training can be applied to datasets related to enterprise resource planning (ERP) and achieve the same or better performance as isolated and centralized training models. “We are now testing an out-of-the-box machine learning model for federated learning and are currently validating this work together with our co-innovation customers,” Li says. He attributes the success of the co-innovation project to strong leadership and open environment provided by SAP Labs China.

REHAU has been live on SAP Intelligent RPA since January 2020, having already replaced two manual processes with the combination of AI and RPA. In one solution, the company was able to reduce its monthly manual processing time of about 1,000 financial accounting documents from four days to 10 minutes. In the other, a bot that combines SAP Intelligent RPA and machine learning reads invoices, extracts the information from them, and then automatically generate payables in SAP S/4HANA Cloud. It can currently process 3,000 product invoices in roughly two minutes.

Chengbo Yu, CIO of the Asia Pacific Region for REHAU Polymer China, is satisfied with the result: “We used SAP Intelligent RPA to automate use cases that required significant manual intervention and redundant work. This saves us time and cuts down on human error, and our employees are spending more time on innovative work.”

Thanks to the following SAP employees for their support: Akie Pan, Camile Zhou, Voga Li, and Cyrano Chen from SAP Labs China; Claire Tseng and Pierre Col from New Ventures and Technologies; Sarah Harvey from Analyst Relations.