Alert fatigue. Blind spots. Constant tuning. Managing alerts in modern observability stacks is harder than it should be. As systems evolve, telemetry changes with your code and business logic — yet alerts are still handcrafted, brittle, and often stale.
In this talk, we’ll explore the common pitfalls of traditional alerting: the lack of semantic consistency across telemetry, the overhead of keeping alerts in sync with evolving systems, and the deluge of low-signal symptoms that drown out meaningful incidents. We’ll show how OpenTelemetry (OTEL) can help standardize signal collection, and more importantly, how it can enable autonomous alerting — where your system continuously adapts, learns from your telemetry, and only raises its hand when it really matters.
Whether you’re drowning in dashboards or just starting your observability journey, this talk offers a vision for how alerting can be smarter, leaner, and less manual
