WEBINAR | Observability in the age of AI agents
Your AI agents are failing.
When a hallucination triggers a production workflow, “it was the model” is not a post-mortem.
This session is for engineering leaders who are done guessing.
ABOUT THE WEBINAR
As organizations pivot from simple chatbots to complex, multi-pass agentic AI systems, traditional observability frameworks are reaching a breaking point. Engineering leaders are no longer just monitoring CPU spikes and latency; they are now accountable for non-deterministic system behaviors, token consumption, and the accuracy of AI-driven decisions.
When an AI agent triggers an automated workflow based on a search result, the cost of “hallucination” isn’t just a bad UI experience, it’s a production incident.
This webinar explores how the OpenSearch Observability Stack provides a unified, vendor-neutral foundation to monitor the entire agent lifecycle. We will move beyond the “data tax” of proprietary vendors and demonstrate how open-source governance enables the transparency, result explainability, and scale required to manage high-stakes AI infrastructure.
Join OpenSearch Software Foundation member representatives Anirudha Jadhav (AWS) and Karsten Schnitter (SAP) as they walk you through exactly how to close that gap using the OpenSearch Observability Stack—a vendor-neutral, open source foundation built for the infrastructure realities of today and tomorrow.
Stop guessing.
Learn how to build a transparent, secure, and vendor-neutral observability foundation for the AI era.
KEY TAKEAWAYS
The shift to agentic health
Why tracking traditional “RED” metrics (Rate, Errors, Duration) is insufficient for AI agents and how to implement confidence signaling and reasoning traces.
Breaking observability silos
Leveraging a single platform to correlate logs, traces, and metrics (including Prometheus integration) to reduce MTTR in distributed AI environments
Sovereignty & compliance
How to maintain data sovereignty and meet strict global compliance standards (GDPR, CRA) by managing sensitive agent telemetry on your own terms.
Cost-effective scalability
Strategies to avoid the “vector refugee” trap, balancing model complexity with ROI-driven infrastructure that scales to petabyte workloads without costly add-ons.
SPEAKERS
Two practitioners who are running production AI infrastructure at scale—not theorizing about it.
Anirudha Jadhav
Senior Engineering Manager, Open Source Observability
AWS
Karsten Schittner
Development Expert
SAP
