Chemicals and materials / Polymer materials
Covestro: Material master data creation and change requests
- Company
- Covestro
- Country
- Germany
- Adoption stage
- In operation
- Source published
- Date basis
- The date the source was published. It can differ from the date adoption started.
- How the source was checked
- Read the full source text
The work problem
At Covestro, business users who needed to create new materials, extend a material's plant assignment, or register a new customer or vendor had no direct way into SAP Master Data Governance (MDG). Users sent emails with no set format, and MDG experts interpreted the requests, clarified attributes, corrected errors, checked for duplicates by hand and then entered the data into the system. According to the AWS blog, a single request could take hours or even days, and the process was split into many steps, was error-prone and depended entirely on people. The AWS blog explained that as the company accelerated its SAP S/4HANA transformation through the ACT program, scaling this manual approach would only increase inefficiency and data quality risk.
Technology and data
Covestro worked with Deloitte Consulting to build two AI agents on AWS: MARIS for material master data and PARIS for business partner master data. Both agents run on Anthropic's Claude models through Amazon Bedrock, and each has its own instructions, tools and session memory. SAP master data is replicated into Amazon S3 through AWS Glue, and Amazon Athena queries material numbers, plant codes, material types and sales organizations in real time to validate inputs. MARIS handles create, change and deletion requests for material master data, including packed material creation, plant extensions and sales organization extensions. Users open MARIS, which is embedded in the Covestro Virtual Assistant (CoVA), in a browser or Microsoft Teams and start a conversation. MARIS guides users step by step to reduce the required input to a handful of essential items, checks for duplicates across multiple systems, and enriches the request with derived data and relevant attributes. Once validation passes, it creates an SAP MDG change request through SAP BTP Integration Suite, and the change request is routed into the standard approval workflow. MDG experts move away from manual corrections and copy and paste work to focus on master data governance. According to the AWS blog, MARIS went live in the legacy SAP ECC environment in early 2026.
Results
According to the AWS blog, the time to process a material master data request fell from an average of 12 hours to 6 minutes, a 99% cut in cycle time. The same post says this applies to about 1,000 material requests a month, or about 12,000 a year. The same post said manual email coordination and handoffs were eliminated entirely. According to the same post, clarification requests that used to fill inboxes fell by 99%. Sunny Liu, Covestro's global process owner for material master data, said the team can now serve stakeholders 24 hours a day, 7 days a week, with no waiting times and no manual dependencies. The AWS blog said PARIS for business partner master data is not yet in production and that quantitative results will be available after its planned go-live on SAP S/4HANA in early 2027.
Limits and open questions
The processing time and clarification request figures come from an AWS blog co-written by Covestro, Deloitte and AWS staff, and the measurement period and comparison method were not disclosed. The exact month MARIS went live and the results of PARIS were not confirmed.
Sources
- Covestro Transforms Master Data Governance with AI Agents on AWSaws.amazon.com, Accessed
Compiled from public sources. These are not results from ATF Works customers.