Bayer Crop Science deploys an agent-based interface to explain results and diagnose errors
Bayer Crop Science has deployed Agentic Decision Engine on top of Connected Decision Engine. The interface uses 22 specialized agents to explain results and diagnose errors in machine learning workflows.
Bayer Crop Science has deployed Agentic Decision Engine as a natural language query interface for Connected Decision Engine, which manages machine learning workflows. Users can investigate the causes of failures, the meaning of results or their origins. An orchestrator agent selects from 22 specialized agents running in production and combines their results for queries spanning multiple areas.
The solution runs on Amazon Bedrock AgentCore and uses LangGraph to orchestrate individual steps. AgentCore Gateway exposes operational system interfaces as MCP tools and controls access through IAM permissions outside the agent code. During diagnostics, a specialized agent links operational logs from Amazon CloudWatch with workflow status from Connected Decision Engine. Answers to questions about specific models also draw on their implementation and documentation.
According to AWS, the solution handles approximately 80 % of user queries without involving technical staff in the initial phase. For the failure diagnostics described, the same case study reports that the time needed to identify the cause fell from more than an hour of manual investigation and waiting for technical support to less than ten minutes.
Why it matters
The case study presents a concrete architecture for bringing documentation, computation status and operational logs together in one interface. Teams managing similar systems can draw on it when designing diagnostics, including how to divide tasks among agents and control their access to tools. The reported results apply to the deployment described.
Two audiences, two different impacts
What this means
For individuals
Breeders, analysts and researchers using this deployment can get explanations of workflow failures without separately reviewing operational logs and finding the right system.
For a business
For Bayer Crop Science, the deployment reduces the need to involve technical staff in routine user queries. Managing permissions outside the agent code also separates access to operational tools from the logic used to generate answers.
ProcessesCheck the original
Event sources
only one source so far · 1 publisher, 0 independent. We count feeds from the same owner only once.