Covestro Deploys MARIS and PARIS AI Agents on AWS Bedrock to Cut Master Data Cycle Time by 99%

Covestro, a global polymer manufacturer headquartered in Leverkusen, Germany, with 48 production sites and approximately 17,500 employees, reduced the time to create master data records from twelve hours to six minutes. The company achieved this by building two AI agents on Amazon Bedrock and integrating them with SAP Master Data Governance (MDG), turning one of its most complex operational processes into a conversational, self-service experience. MARIS (Material Artificial Realtime Intelligence Specialist) handles create, change, and deletion requests for material master data, including packed material creation, plant extensions, and sales organization extensions. MARIS went live in Covestro's legacy SAP ERP (ECC) environment in early 2026. PARIS (Partner Artificial Realtime Intelligence Specialist), planned for go-live on SAP S/4HANA in early 2027, specializes in business partner data management with built-in duplicate detection, address validation, and golden record enrichment using CDQ technology cross-referenced against commercial registers and premium sources such as Dun & Bradstreet. The transformation delivers a 99% reduction in cycle time for approximately 1,000 material requests per month, or roughly 12,000 per year. The architecture is layered and event-driven, with users interacting through a chatbot on AWS Amplify or Microsoft Teams, requests flowing through Amazon API Gateway to AWS Lambda hosting the AI agent runtime, and a Supervisor Agent built with the Strands Agents SDK on Amazon Bedrock AgentCore classifying intent and routing to specialized task agents running on Anthropic Claude models. SAP master data is replicated via AWS Glue into Amazon S3 as Data Mesh tables, with Amazon Athena validating inputs before calling SAP APIs through SAP BTP Integration Suite. Sunny Liu, Global Process Owner for Material Master Data at Covestro, noted that the agent has provided instant, conversational access to complex processes, with stakeholders now served 24/7 with no waiting times and no manual dependencies.