How to build smarter OpenSearch alerts: Join our live conversation
The New Stack Charles Humble
OpenSearch is adding smarter alerts and a central Alert Manager. It’s built for noisy AI-era observability, where false positives eat on-call time.
Based on reporting by The New Stack, Charles Humble — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenSearch is getting a pair of new alerting tools aimed squarely at the mess modern observability has become. On September 10, Joshua Bright, a senior product manager at AWS OpenSearch, will walk SREs and platform engineers through them in a technical session built around real-world alerting pain.
The pitch is simple: teams already have plenty of telemetry, but the tools used to act on it still lag behind. That problem gets sharper as AI agents add more signals to logs, metrics, traces, and dashboards. The Linux Foundation says 77% of organizations already treat OpenSearch as a core or supporting part of their AI infrastructure, and agent tracing is one reason why.
The first new piece is Piped Processing Language, or PPL, for alerting. It uses a Unix-style pipeline model so engineers can filter, transform, and correlate data across logs, metrics, and traces with readable syntax. Instead of relying on single-threshold rules, teams can chain conditions together and catch more subtle failures, like a latency spike in an AI agent’s tool call that only matters when error rates in logs rise too.
The second piece is a unified Alert Manager. It gives teams one place to handle alert rules, routing, suppression, and escalation, which is a lot closer to how incidents are actually handled when things go wrong. The company says the webinar will include a live demo against a realistic observability workload, plus an open Q&A with the Amazon team.
There’s also a licensing angle here. Everything in the webinar ships under Apache 2.0, with no feature gating. That’s the kind of detail open-source buyers notice when they’ve been burned by half-open platforms that charge extra the moment the useful part appears.
My take — AI-written commentary, not fact-checked reporting
OpenSearch is making the right bet here: observability tools are drowning in their own alert noise, and AI agents only make the flood louder. Centralizing routing and making queries easier to reason about is more useful than yet another dashboard with heroic marketing copy. The industry keeps pretending alerting is a side feature, and then acts surprised when on-call turns into a lottery.
Read more about this at: The New Stack