How Cornerstone OnDemand cut database diagnosis by 78% with Amazon Bedrock
aws ·
Cornerstone OnDemand has significantly improved its database operations by implementing Orion AI, a multi-agent system built on Amazon Bedrock and Strands Agents. This solution reduced the average time required for database diagnosis from 45 minutes to just 10 minutes, representing a 78% efficiency gain. The system automates complex workflows that previously involved over ten manual steps, condensing them into single natural language interactions. Additionally, Orion AI eliminated the reporting lag between site reliability engineering and data teams, providing instantaneous updates instead of a 15-minute delay. The platform also reduced redundant alert noise by 65% through advanced deduplication and threshold filtering mechanisms. Designed with a hub-and-spoke topology, the system uses specialized agents for distinct operational domains rather than relying on a single general-purpose model. Key design principles include hybrid routing for speed and accuracy, as well as bypassing conversational memory for real-time metrics to ensure diagnostic freshness. A three-person team delivered this comprehensive solution in six months, demonstrating the viability of agentic AI in enterprise DataOps environments.
aws