Observability has always been a paradox. Platform engineering teams invest heavily in logs, metrics, and traces, yet when an outage hits, they still scramble to find root causes. The data exists, but connecting it in real time is slow, expensive, and often manual.
At KubeCon 2025, Observe Inc. unveiled two new AI-driven agents, AI SRE and o11y.ai, designed to fix that disconnect. Built on Observe’s data lake architecture and knowledge graph, these agents transform observability from a passive data store into an active, intelligent assistant that helps engineers solve problems faster and at lower cost.
The early numbers are hard to ignore:
- Incident triage up to 10x faster
- Mean time to resolution (MTTR) reduced from hours to minutes
- Observability costs down by as much as 60%
For SREs, these aren’t incremental gains; they’re a rethinking of what observability can do.
AI-Powered Reliability: From Reactive to Predictive
The idea behind Observe’s new agents is simple but powerful: use AI to do the heavy lifting of correlation and triage so engineers can focus on solutions instead of searching.
AI SRE acts as a digital teammate that understands the shape of production systems. It maps signals across logs, metrics, and traces using Observe’s knowledge graph, automatically spotting patterns and suggesting likely root causes. It can detect that a latency spike isn’t just “another alert”; it’s a cascading timeout caused by a specific downstream service or config drift.
Meanwhile, o11y.ai focuses on developer workflows. It connects directly with CI/CD and OpenTelemetry pipelines, turning observability data into actionable context during deployment and post-release debugging. Developers can ask natural language questions such as “What changed right before this error rate jumped?” and get precise, contextual answers drawn from the live system graph.
The result: less context switching, faster fixes, and a smoother handoff between SREs and developers.
Why This Matters Now
Kubernetes and microservices have multiplied the complexity of production environments. Each release ships hundreds of components and dependencies, generating billions of telemetry events every day. Traditional observability tools treat that flood of data as something to store and query. Observe treats it as something to reason about.
By embedding AI at the SRE layer, Observe shifts observability from reactive monitoring to proactive reliability engineering. The system doesn’t just tell you what happened; it helps you understand why and how to fix it.
This comes at a time when most organizations are struggling with observability’s runaway costs. Centralized telemetry pipelines can consume more budget than the applications they monitor. Observe’s agents tackle that problem at the source by cutting down data duplication and automating the most expensive step of all: human analysis.
For Practitioners: Faster Insights, Lower Noise
For the platform engineers on call, this means fewer late nights chasing phantom alerts. AI SRE can spot redundant notifications and automatically correlate related issues before escalating. It effectively “de-noises” observability by clustering signals that stem from the same root cause.
For developers, o11y.ai turns telemetry into a true feedback loop. Instead of sifting through dashboards, they can query system behavior in plain language, backed by context from recent code changes, deployment data, and infrastructure events.
And for platform teams, the financial benefit is clear: the same telemetry infrastructure now delivers more value per byte collected, with automated summarization and correlation reducing the need for high-volume ingestion.
The Takeaway
For SREs and platform engineers, Observe’s AI SRE and o11y.ai agents mark a shift from tool-centric observability to outcome-centric reliability.
They don’t just make monitoring smarter; they make operations calmer. They don’t just speed up debugging; they close the loop between detection and recovery. And they do it all on the foundation of open data, not black-box automation.
As observability enters the AI era, Observe’s message is simple but timely: It’s time for your tools to think with you, not just for you.
