Orion AI: How Cornerstone OnDemand Cut Database Diagnosis by 78% with a Three-Agent System on Amazon Bedrock

Description and use

Cornerstone OnDemand implemented Orion AI, a three‑agent system on Amazon Bedrock, reducing database diagnosis time from 45 to 10 minutes, automating 60 % of incidents, and handling over 40 alerts daily.

Problem

Manual database diagnosis at scale took an average of 45 minutes per alert, consuming most engineers’ time on repetitive tasks.

Outcome

In six months, Orion AI reduced diagnosis time by 78 %, automated 60 % of incidents, and enabled handling of 40+ alerts daily with structured audit trails and approval gates.

Workflow

• 1. Alert ingestion via EventBridge
• 2. Diagnosis Agent analyzes alert data and metrics
• 3. Query Generation Agent creates safe, read‑only SQL and validates syntax
• 4. Executed queries via Lambda with safety checks
• 5. Remediation Agent proposes actions and requires human approval for destructive ops
• 6. Approved fixes executed within transaction boundaries
• 7. All steps logged for audit and observability

Evidence

Cornerstone OnDemand built Orion AI to turn database operations from reactive firefighting into proactive automation.

Source

A three-person team cut database diagnosis from 45 minutes to 10 minutes in six months using Amazon Bedrock and the Strands Agents framework.

Source

Diagnosis time dropped from 45 minutes to 10 minutes (78% reduction)

Source