Agentic Observability — The Path to AI Co-SRE
summary
Claims specific reproducible results: a 18-model consensus engine (theoretical FPR 10⁻³⁶) reduces false positives, a 22 specialized troubleshooting agents composable framework enforces schemaed worker outputs and automatic validators, and Netdata runs per-metric, per-second ML at the edge with models trained every 3 hours (18 overlapping windows). Built systems find real faults (auditd misconfiguration that panicked servers) and remain limited on short-lived workloads, crashed services with no logs, and extremely slow drifts.
AI-written summary · about
chapters (29)
- 0:00Agentic observability intro
- 1:31AI struggles with raw data
- 2:32Netdata ML anomaly detection
- 4:16Hierarchical anomaly tree
- 5:38Machine learning reveals hidden infrastructure correlations
- 7:45Netdata learns system patterns
- 8:30Ensemble models cut false positives
- 10:00Anomaly types Netdata detects
- 11:02Current detection limits
- 12:24Single-agent AI fails RCA
- 14:30Disciplined AI investigators design
- 16:37Enforced investigation workflows
- 18:43Worker agents and tasks
- 20:50Validate worker reports auto
- 21:55Case study: year-long reboot mystery
- 24:19How agents establish baselines before and after incidents
- 26:06From reactive to proactive AI
- 27:54AI auto-generated reports
- 28:36Live DB insights visualization
- 30:22Future proactive monitoring
- 31:06Collect all observability data
- 32:33Auto dashboards high-res data
- 33:35Per-second observability magic
- 34:17Start agentic tools guide
- 36:20Path to senior SRE with AI
- 38:07Model selection for AI investigations
- 39:54Cost efficiency in AI co-SRE
- 41:17Netdata cuts observability costs
- 43:25Symptoms vs root causes
related talks
28mDO WE STILL NEED TO "OBSERVE" THE FUTURE OF AI & OBSERVABILITY
32mSTOP TREATING AI SRE AS JUST AN ENGINEERING PROBLEM
40mWhen Incidents Fix Themselves: AI SRE in action
41mAIRe - AI Reliability Engineering
41mGenerative AI: bliss or miss?
11mSRE Challenges and Solutions for High-Availability CCAI Systems
43mCONTEXT ENGINEERING FOR SELF-HEALING AI SRE
50mAI IN SRE: HYPE VS REALITY IN 2026