Skip to content

Agentic AI#

Concepts#

Agentic AI in Unryo refers to the use of intelligent agents powered by Large Language Models (LLMs) and domain-specific reasoning.
These agents operate on top of Unryo’s topology, metrics, events, logs, and monitoring stack to automate:

  • Issue detection
  • Root-cause analysis
  • Remediation recommendations or actions

Agentic AI bridges existing observability tools with AI-powered insights and automation.

Key Components#

Component Purpose
Unified Topology Engine Provides real-time dependency mapping across infrastructure, cloud, network, and applications. Serves as the context foundation for AI agents.
Multi-LLM Support Works with multiple LLM providers, including private/self-hosted models. Provides flexibility and data governance options.
Data Integration Connects to metrics, logs, events, CMDBs, and external monitoring tools. Creates a rich and correlated context.
AI Agents & Automation Detect anomalies, generate hypotheses, test probable root causes, and suggest or initiate remediation.
AI Assistant Interface Natural-language interaction: users can ask “Why is my app slow?” and receive contextual answers with impact analysis and recommended actions.

Workflow#

The Agentic AI workflow extends Unryo’s correlation engine:

  1. Data Ingestion
    Connects to existing monitoring sources. Data is normalized, enriched, and linked to live topology.

  2. Topology Mapping
    Builds a real-time map of dependencies: networks, hosts, storage, applications, and services.

  3. Anomaly Detection
    Identifies deviations, threshold breaches, or abnormal patterns across metrics, logs, and events.

  4. Correlation & Root Cause
    Uses topology and historical data to distinguish between root causes and secondary impacts.

  5. Impact Mapping
    Determines which business services, customers, or SLAs are affected.

  6. AI-Assisted Insights
    Surfaces context, probable causes, relevant logs/metrics, and remediation options. Can trigger automated actions if enabled.

  7. User Interaction
    Operators interact via natural language, drill into past incidents, compare patterns, or escalate with enriched context.

  8. Continuous Learning
    Feedback loops improve accuracy over time, reducing false positives and adapting to evolving infrastructure.

Relationship with Correlation Workflow#

Agentic AI builds on Unryo’s correlation foundation by adding:

  • Automated hypothesis generation and diagnostics
  • Natural-language explanations and summaries
  • Remediation suggestions or automation triggers
  • Cross-tool, cross-silo correlation for end-to-end visibility

Configuration#

To enable Agentic AI:

  • Select an LLM provider (cloud-based or self-hosted).
  • Connect monitoring sources: metrics, logs, events, CMDBs, ticketing tools.
  • Enable topology discovery or import external topology.
  • Configure anomaly detection thresholds and correlation rules.
  • Define permissions for AI Assistant access.

Benefits#

  • Faster MTTR – accelerated troubleshooting and resolution.
  • Noise Reduction – suppresses redundant alerts, focuses on root causes.
  • Cross-Silo Visibility – unifies data across tools, domains, and services.
  • Decision Support – contextual, AI-driven insights for operators.
  • Scalable Automation – supports large and dynamic environments without proportional staffing increases.