Sentinel AI

Sentinel AI is an AI-powered SRE that analyzes production incidents, identifies root causes, estimates business impact, and delivers actionable fixes in seconds.

Comment

Inspiration

Modern production incidents often take hours to investigate because engineers must manually correlate logs, metrics, traces, and architecture diagrams across multiple tools. During critical outages, valuable time is lost switching between dashboards, searching documentation, and validating hypotheses. We wanted to build an AI-powered engineering command center that acts like an experienced Level 3 Site Reliability Engineer (SRE), enabling teams to identify root causes and recover from incidents in minutes instead of hours.

What it does

Sentinel AI is an autonomous incident investigation platform that helps DevOps, SRE, and engineering teams resolve production failures faster. Users upload production logs, metrics, traces, and architecture diagrams. Sentinel's AEGIS Engine orchestrates multiple specialized AI agents that analyze the evidence in parallel. The platform streams live investigation progress, identifies the most likely root cause, estimates business impact, detects affected services, and generates a comprehensive remediation plan with a postmortem report.

Key features include:

Multi-agent AI incident investigation

  • Real-time investigation progress via WebSockets
  • AI-generated root cause analysis
  • Business impact assessment
  • Actionable remediation recommendations
  • Automated incident postmortem generation
  • Interactive engineering dashboard

How we built it

We built Sentinel AI using a modern full-stack architecture.

Frontend

  • React.js
  • Vite
  • React Router
  • Tailwind CSS
  • Framer Motion
  • Recharts

Backend

  • Node.js *Express.js *PostgreSQL (Neon) *JWT Authentication
  • Multer for file uploads *WebSockets (Socket.IO)

AI Layer

  • OpenAI GPT models
  • Multi-agent orchestration (AEGIS Engine)
  • Parallel evidence analysis
  • AI-generated summaries and remediation

Deployment

  • Vercel (Frontend)
  • Render (Backend)
  • Neon PostgreSQL
  • GitHub for version control

Challenges we ran into

One of our biggest challenges was designing an architecture that could simulate multiple AI specialists working together instead of relying on a single LLM response. We also had to build a responsive real-time investigation interface using WebSockets while ensuring the user experience remained smooth.

On the deployment side, we resolved several production issues, including CORS configuration, Vercel routing for React applications, backend deployment on Render, environment variable management, and secure API communication between services.

Another challenge was transforming raw infrastructure data into concise, actionable insights rather than overwhelming users with AI-generated text.

Accomplishments that we're proud of

  • Built a complete end-to-end AI-powered incident investigation platform.
  • Successfully implemented a multi-agent AI workflow for production incident analysis.
  • Created a modern engineering dashboard with live investigation updates.
  • Automated root cause analysis, business impact assessment, and postmortem generation.
  • Deployed the application using a scalable cloud architecture with Vercel, Render, and Neon PostgreSQL.
  • Designed a solution that demonstrates how AI can significantly reduce incident response time for engineering teams.

What we learned

Throughout this project, we gained valuable experience in building production-ready AI applications, orchestrating multiple AI agents, integrating real-time communication with WebSockets, and deploying distributed full-stack systems in the cloud.

We also learned the importance of prompt engineering, structured AI workflows, secure authentication, scalable backend architecture, and designing user experiences that present complex technical insights in a simple and actionable way.

What's next for Sentinel AI

Our vision is to evolve Sentinel AI into a fully autonomous engineering reliability platform.

Our roadmap includes:

  • Integration with Kubernetes, Docker, AWS CloudWatch, Azure Monitor, Datadog, Grafana, Prometheus, and Splunk.
  • Automatic incident detection without manual uploads.
  • CI/CD pipeline integration with GitHub Actions and GitLab.
  • Slack, Microsoft Teams, and PagerDuty integrations for real-time collaboration.
  • Historical incident learning and AI-powered preventive recommendations.
  • Predictive failure analysis using machine learning.
  • AI-powered infrastructure optimization and cost recommendations.
  • Enterprise-grade RBAC, audit logs, compliance, and multi-tenant support.
  • Support for on-premises and hybrid cloud deployments.

Ultimately, we envision Sentinel AI becoming the autonomous AI engineering teammate that helps organizations proactively detect, investigate, and resolve production incidents before they impact customers.

Share this project:

Updates