Network Guardian

AI-Driven Multi-Agent System for Network Anomaly Detection and Diagnosis

Pilot Overview

Network Guardian is an AI-driven multi-agent system that leverages LangGraph and Long-Term Memory (LTM) to identify and diagnose network anomalies. This pilot phase demonstrates the feasibility and potential value of this approach using synthetic data in a controlled environment.

Key Features
  • Multi-agent collaboration using LangGraph orchestration
  • Semantic, episodic, and procedural memory for intelligent reasoning
  • Anomaly detection and root cause diagnosis
  • Clear presentation of findings for human evaluators

Agent System Architecture

Monitor Agent Anomaly Detection Triage Agent Event Correlation Diagnosis Agent Root Cause Analysis Human Agent Result Presentation Semantic Memory Episodic Memory Procedural Memory
Agent Roles
  • Monitor Agent: Detects anomalies in network metrics and logs
  • Triage & Correlation Agent: Correlates related events and prioritizes issues
  • Diagnosis Agent: Determines root causes using RAG and memory systems
  • Human Interaction Agent: Presents findings clearly for human review
Memory Systems
  • Semantic Memory: Network topology, component knowledge, baselines
  • Episodic Memory: Historical events and past incidents
  • Procedural Memory: Diagnostic procedures and resolution patterns

Getting Started

To begin exploring Network Guardian's capabilities:

  1. Visit the Dashboard to view the synthetic network topology and status
  2. Go to the Analysis page to run the multi-agent system on predefined synthetic datasets
  3. Review the generated diagnoses and agent reasoning
  4. Provide feedback to help improve the system
Go to Dashboard