Why I built my own monitoring tool instead of using the enterprise one

When you manage contact center operations across multiple sites and hundreds of agents, real-time data isn't a nice-to-have — it's the foundation of every decision you make.

The enterprise monitoring tools we had were built for dashboards in boardrooms, not for operations teams making split-second decisions. Updates lagged 30+ seconds behind reality. Ghost agents cluttered the views. WebRTC drops went undetected until customers started complaining.

So I built MONTI.

MONTI is a real-time monitoring application built on event-driven architecture. Instead of polling APIs every 30 seconds like the enterprise tools, it subscribes to live events and delivers sub-second updates. It knows when an agent drops a WebRTC connection before the agent even realizes it. It detects queue anomalies that the built-in tools simply miss.

The stack is intentionally lean: Go on the backend for concurrency and performance, Vite + React on the frontend for a responsive UI, WebSockets for real-time data delivery, and the whole thing deployed on AWS with Terraform, monitored through Grafana and Prometheus, with Keycloak handling auth.

What surprised me most wasn't the technical challenge — it was how quickly a purpose-built tool outperformed a platform backed by millions in development budget. The difference? I understood the operational domain. I knew exactly what data mattered and how it needed to be presented.

That's the lesson: domain knowledge isn't separate from engineering skill. It IS engineering skill.