AgenticOps: AI Agent-Based Autonomous Operations
Reading time: About 2 minutes
AgenticOps is an approach to autonomously build a feedback loop through AI agents for continuous improvement in production environments after developing software with AIDLC. While traditional AIOps used AI as a monitoring aid, AgenticOps enables AI agents to autonomously perform detection → decision → execution based on observability data.
Relationship with AIDLC
If AIDLC focuses on "how to build" (development methodology), AgenticOps focuses on "how to operate and improve" (operational feedback loop). Domain constraints defined by AIDLC's ontology are used as criteria for operational decisions by AgenticOps AI agents, and insights discovered during operations are fed back as the Outer Loop for ontology evolution.
Structure
Reading in the order 1 → 2 → 3 allows you to follow the entire journey from data-based construction to autonomous operations realization.
| Order | Document | Core Question |
|---|---|---|
| 1 | Observability Stack | How do we collect and analyze operational data? |
| 2 | Predictive Operations | How do we predict and prevent failures in advance? |
| 3 | Autonomous Response | How do AI agents respond autonomously? |
Core Foundation: AWS Open Source Strategy
AWS provides core Kubernetes ecosystem tools as Managed Add-ons (22+) and managed open source services (AMP, AMG, ADOT). On this foundation, Kiro + MCP (Model Context Protocol) operates as the core tool for AgenticOps, autonomously controlling EKS clusters, analyzing CloudWatch metrics, and optimizing costs through AWS MCP servers (50+ GA).