AI Engineering Assistant

A Multi-Agent AI System for Production Incident Investigation and Root Cause Analysis

Project Overview

This project is a proof-of-concept demonstrating a multi-agent AI system for engineering incident investigation. Specialized agents collaborate to analyze production incidents, logs, deployment information, and metrics to identify likely root causes and recommend solutions.

Features

• Multi-Agent Workflow — Coordinated agents work together to investigate incidents end to end
• Incident Investigation — Structured analysis of production incidents and their context
• Log Analysis — Automated review of application and system logs
• Metric Analysis — Evaluation of performance and health metrics during incidents
• Root Cause Analysis — Identification of probable underlying causes
• AI-Powered Recommendations — Actionable suggestions for remediation and prevention
• Retrieval-Augmented Generation (planned) — Context-aware responses grounded in operational knowledge
• REST API — Programmatic access to investigation workflows and results

High-Level Architecture

User Question
        |
        V
Orchestrator
        |
--------------------------------------------
|      |       |        |        |         |
Retrieval Log  Metric   RCA   Recommendation Reviewer
 Agent   Agent Agent    Agent      Agent       Agent
        |
        V
Final Report

The orchestrator receives a user question and delegates work to specialized agents. Each agent contributes its analysis, and the workflow culminates in a consolidated final report.

Technology Stack

• Java 21
• Spring Boot 3
• Maven
• Spring AI (planned)
• OpenAI (planned)
• REST API

Project Status

This project is being developed incrementally. Each Git commit introduces one new capability, building toward a complete multi-agent incident investigation system.