How agentic telemetry cuts observability cost and improves signal

Most engineering teams send far more telemetry than they actually use, driving up costs while degrading signal. And the problem is accelerating as AI coding agents generate even more instrumentation, scaling telemetry without enforcing standards and increasing the burden on DevOps to manage the resulting noise, cost, and inconsistency downstream.

Trading in your observability backend for a cheaper solution might seem like the answer, but switching is an expensive process and doesn’t fix the root cause. The solution lies in managing how telemetry is generated and controlled throughout the entire lifecycle, from instrumentation at code all the way into runtime. Join Sawmills CTO Amir Jakoby and CPO Erez Rusovsky for a session on how teams are using agentic telemetry management to reduce observability cost and improve signal.

You’ll learn how to manage telemetry across the lifecycle by:

• Operating telemetry in-flight, analyzing production data to identify waste and quality issues like verbose logs, duplicate patterns, high-cardinality metrics, and schema violations, then recommends fixes that can be applied directly to your pipeline.

• Closing the loop from production back to code and CI by using runtime insights to help developers catch telemetry issues before they ship, standardize instrumentation, and prevent instrumentation drift earlier in the lifecycle.

• Giving developers telemetry autonomy within DevOps guardrails by surfacing ready-to-apply recommendations in Slack, Teams, or the Sawmills UI, while DevOps maintains visibility, approval and rollback control.

• Improving observability without replacing your backend by adding an intelligent control layer before ingestion, so the data flowing into Datadog, Splunk, New Relic, or any existing platform is cleaner, more useful, and worth paying for.

We look forward to seeing you there!

Questions? Contact [email protected]