Agentic AI for observability and troubleshooting with Amazon OpenSearch Service

This article introduces three new agentic AI features in Amazon OpenSearch Service designed to accelerate incident troubleshooting and reduce Mean Time to Resolution (MTTR).

• Agentic Chatbot understands page context and generates queries without manual coding expertise
• Investigation Agent performs hypothesis-driven analysis across multiple data indices automatically
• Agentic Memory improves accuracy and speed through persistent learning across interactions
• Features accessible via "Ask AI" button in OpenSearch UI with no additional infrastructure required
• Investigation Agent correlates data across indices using trace IDs to identify root causes
• Powered by Claude Sonnet 4.6 model in Amazon Bedrock with full explainability
• Available at no cost with optional customer-managed encryption for investigation results

These capabilities enable engineers to transition from alerts to root cause identification in minutes, automating complex multi-index correlation and analysis that previously required manual expertise.