DevStream – AI-Powered Incident Monitoring Pipeline

AI-powered incident monitoring pipeline built with n8n, Gemini AI, and PostgreSQL to automate API failure analysis and engineer notifications.

Project Overview

DevStream is an AI-powered incident monitoring pipeline built using n8n, Gemini AI, and Neon PostgreSQL.

The project automates API failure handling by receiving failure events, validating requests, calculating incident severity, generating AI-assisted root cause analysis, storing incident data in PostgreSQL, and notifying engineers about high-severity incidents.

It demonstrates workflow automation, AI integration, database persistence, and event-driven incident management in a backend engineering scenario.

Project Flow

API Failure Event
      │
      ▼
Webhook
      │
      ▼
Payload Validation
      │
      ▼
Severity Calculation
      │
      ▼
PostgreSQL Failure Log
      │
      ▼
Gemini AI Analysis
      │
      ▼
Store AI Analysis
      │
      ▼
HIGH Severity?
      │
      ▼
Engineer Notification

Features

• AI-assisted root cause analysis using Gemini AI
• Automated incident severity calculation
• PostgreSQL persistence for API failures and AI analysis
• Conditional email notifications for HIGH-severity incidents
• Workflow automation using n8n
• Structured storage of AI-generated analysis and recommendations
• Extensible architecture for future Spring Boot integration

Tech Stack

Category	Technology
Workflow Automation	n8n
AI Model	Gemini AI
Database	Neon PostgreSQL
Database Engine	PostgreSQL
Notification	Gmail
API Testing	Postman
Version Control	GitHub
Future Event Source	Spring Boot Microservice

Architecture Diagram

flowchart TD

A["API Failure Event"]
B["Receive Failure Event"]
C["Validate Payload"]
D["Calculate Severity"]
E["Log API Failure"]
F["Analyze Failure (Gemini AI)"]
G["Parse AI Response"]
H["Store AI Analysis"]
I["Update AI Status"]
J{"HIGH Severity?"}
K["Notify Engineer (Email)"]
L["End"]
M["Neon PostgreSQL
(api_failure_logs, ai_analysis)"]

A --> B
B --> C
C --> D
D --> E

E -.->|"Save Failure Log"| M

E --> F
F --> G
G --> H

H -.->|"Save AI Analysis"| M

H --> I
I -.->|"Update Processing Status"| M
I --> J

J -->|Yes| K
J -->|No| L

Loading

Current Event Source: Postman for API failure simulation

Workflow Trigger: n8n HTTP Webhook

Planned Event Source: Spring Boot Microservice (Phase 2)

Workflow Screenshot

The following screenshot shows the complete implementation of the AI-powered incident monitoring pipeline in n8n.

The workflow validates incoming API failures, performs AI-assisted incident analysis, persists the results in Neon PostgreSQL, and notifies engineers for HIGH-severity incidents.

Demo

Sample Workflow Execution

1. API failure event is sent using Postman.
2. n8n validates the payload and calculates severity.
3. Failure details are stored in Neon PostgreSQL.
4. Gemini AI analyzes the incident and generates:
   • Root cause
   • Java fix recommendation
   • Unit test suggestion
   • Best practice
   • Confidence score
5. AI analysis is stored in PostgreSQL.
6. HIGH severity incidents trigger an automated email notification to the designated engineer.

Database Schema

The project uses Neon PostgreSQL to store both raw API failure events and AI-generated incident analysis.

The following diagram presents a simplified view of the primary database fields and relationships.

The database consists of two primary tables:

Table	Purpose
api_failure_logs	Stores API failure details received by the workflow
ai_analysis	Stores AI-generated analysis linked to each API failure

erDiagram

api_failure_logs ||--o| ai_analysis : has

api_failure_logs {
    int failure_id PK
    string service_name
    string endpoint
    int status_code
    string severity
    string ai_status
}

ai_analysis {
    int analysis_id PK
    int failure_id FK
    string root_cause
    string java_fix
    string recommended_action
    float confidence_score
}

Loading

Each API failure is stored first in api_failure_logs.

After AI processing, the generated root cause analysis, Java fix recommendations, confidence score, and other outputs are stored in ai_analysis using the corresponding failure_id.

Sample API Payload

The workflow receives API failure events through an HTTP Webhook.

Example request:

{
  "serviceName": "Order Service",
  "endpoint": "/api/orders",
  "httpMethod": "POST",
  "statusCode": 500,
  "responseTimeMs": 2400,
  "errorMessage": "NullPointerException while creating order",
  "stackTrace": "java.lang.NullPointerException..."
}

Setup Guide

Prerequisites

• n8n Cloud or self-hosted n8n
• Google Gemini API Key
• Neon PostgreSQL
• Gmail Account
• Postman

Installation

1. Clone the repository.
2. Create the PostgreSQL tables using the SQL scripts.
3. Import the n8n workflow JSON.
4. Configure the Gemini API credentials.
5. Configure PostgreSQL credentials.
6. Configure Gmail credentials.
7. Activate the workflow.
8. Send the sample payload using Postman.
9. Verify the AI analysis and database entries.

Security

This repository intentionally excludes all secrets and sensitive credentials.

The exported n8n workflow contains only references to credentials—not the actual values.

After importing the workflow, configure your own credentials in n8n for:

• Google Gemini API
• Neon PostgreSQL
• Gmail OAuth

Never commit:

• API keys
• Database passwords
• Connection strings
• OAuth tokens
• .env files
• Exported credentials

Future Enhancements

Phase 2

• Replace Postman with Spring Boot Microservice
• Automatic failure reporting
• Exception handling integration

Planned Improvements

• Slack notifications
• Jira integration
• Monitoring dashboard
• Incident analytics
• Retry recommendations
• Historical incident search
• Docker deployment

Project Structure

DevStream/
│
├── README.md
├── workflow/
│   └── DevStream_AI_Incident_Monitoring.json
├── database/
│   ├── create_tables.sql
│   └── sample_data.sql
├── images/
│   └── workflow.png
└── LICENSE

Project Status

Current Phase

Phase 1 – AI-powered incident monitoring workflow using Postman for event simulation.

Next Phase

Integrate a Spring Boot microservice to automatically report API failures to the n8n workflow.

Repository Contents

• workflow/ – n8n workflow export
• database/ – PostgreSQL schema and sample data
• images/ – Architecture and workflow screenshots
• README.md – Project documentation

License

This project is licensed under the MIT License. See the LICENSE file for details.