Covestro Unlocks Internal Research Insights with Mergeflow Software Using AWS

Using software from AWS Partner Mergeflow, Covestro—a global company that produces precursors for high performance materials—can now find relevant research and development (R&D) insights with significantly higher efficiency than it previously took to conduct research. The tool uses advanced artificial intelligence (AI), natural language processing, and other technologies to extract insights from both internal and external data sources. It uses Amazon EC2 for secure and resizable compute capacity, and Amazon EBS for easy-to-use, high-performance block storage at any scale.

German chemical company Covestro—merged from Bayer in 2015 and previously called Bayer MaterialScience—is a global company that produces a wide range of innovative raw materials for high-performance materials and plastics. It employs 18,000 people, including about 1,500 research and development (R&D) scientists who work to create innovative new products and materials. Because of this, employees have a critical need to access up-to-date R&D and market information from around the world.

For several years now, Covestro has been using software from AWS Partner Mergeflow, to search, gather, and analyze external data for its R&D efforts. But its legacy IT infrastructure meant it couldn’t use Mergeflow’s software to apply those same research capabilities to Covestro’s extensive body of in-house research. That changed when Covestro began migrating its IT systems to Amazon Web Services (AWS). As the migration got underway, both Covestro and Mergeflow saw an opportunity to extend Mergeflow’s tool to Covestro’s internal datasets and data repositories.

Covestro’s researchers can now search both internal and external information sources from a single interface using Mergeflow’s software. This has eliminated data silos, reducing the amount of time and effort it previously took researchers to find relevant information. And recent advances in artificial intelligence (AI) capabilities make searches easier than ever, essentially giving users research “superpowers,” according to Florian Wolf, chief executive officer (CEO) of Mergeflow.

Paul Heinz, Covestro’s head of research and development digital processes, says his organization has gained great insights since it first began using Mergeflow’s software to find and analyze data from external sources. Those sources include worldwide patent offices, scientific publications, research databases, technology transfer documents, government-funded research projects, venture capital news, and market analyses. The company had explored applying Mergeflow’s tool to its internal information streams, but was unable to execute it efficiently because of the technical limitations of its IT infrastructure. That changed when Covestro migrated to AWS.

The migration made it possible to create a private cloud environment on AWS through which Mergeflow’s software can search Covestro’s internal research and data repositories. Doing this eliminates the information silos that previously made it hard to search for information distributed across the globe.

Wolf describes the challenges of researching Covestro’s internal data before the migration. “You might, for example, be looking for someone who you could ask about a certain technology,” he says. “But you can’t use the company directory because it doesn’t have the information about what the person is working on.” As a result, finding relevant internal research and the right colleagues to speak with sometimes involved more time and effort than researching external data did.

As Covestro began migrating its systems to AWS, Mergeflow began deploying its software in a modular fashion to Covestro’s internal data sources. It uses Amazon Elastic Compute Cloud (Amazon EC2) for secure and resizable compute capacity for virtually any workload, along with Amazon Elastic Block Store (Amazon EBS) for easy-to-use, high-performance block storage at any scale. It also uses Amazon Aurora, a relational database management system built for the cloud, and Amazon Simple Storage Service (Amazon S3) for object storage built to retrieve any amount of data from anywhere.

Managing this deployment during the height of the COVID-19 pandemic was challenging, because both Covestro’s and Mergeflow’s teams had to work remotely. But Heinz said the two organizations had excellent communication that made it possible to work together closely and effectively. “It became so normal to us to just work remotely that, at some point, I didn’t even perceive it as remarkable,” he says. “But it was totally remarkable.”

As the project progressed, Covestro was able to begin using Knowledge Explorer—the name for the internal deployment of Mergeflow’s software—to search its many internal data repositories. Because the system brings together disparate external and internal data sources in one user interface, employees can perform searches much more quickly—in some instances, 90 percent faster than it used to take. The system was built using Covestro’s existing user access management system, so employees didn’t need to learn a new authentication system or remember new usernames, passwords, or other credentials.

Since it became available, Knowledge Explorer has completely changed how Covestro employees look for, analyze, and use research data. “Just imagine—you place one query into Knowledge Explorer and receive results from tens of different datasets, and each of these datasets includes tens or hundreds of different sources,” says Heinz. “In the past, for all of these datasets, people had to use different tools. This increases the creative part of their scientific work, but it also allows people to save a lot of time.”

The addition of powerful new AI capabilities has helped researchers gain even more benefits, adds Wolf. “You can do things that before were completely unthinkable,” he says. “It’s helping you understand unstructured content in ways that were previously not possible.”

By eliminating time-consuming manual searches and acting as more of a research partner that users can communicate with using natural language queries, the AI-powered system is also making research more interesting and fun, Wolf adds. “It’s really about giving people superpowers, I think, if you do it right,” he says. Heinz adds that he expects Covestro will discover even more benefits as the tool’s AI capabilities become more advanced. “I personally believe that over the next few years, we will build even more than we currently envision,” he says.