AgentOps

AgentOps: The missing AI governance layer for AI. We provide real-time telemetry, immutable audit logs, strict cost enforcement, and human in loop safeguards for autonomous agents.

Comment

The Inspiration:

Everyone is rushing to deploy autonomous ai agents but practically, no one is building the brakes. People and Enterprises are are handing over API keys and just hoping that it don't burn through thousands of dollars overnight. AgentOps is a platform where human can monitor, pause, and audit autonomous agents in real time.

How I Built It:

Being a BCA student and parallelly building projects for hackathons and participating in competitions I prefer vibe coding and fast prototyping.

I Vibe Coded the platform using v0.app and used Next.js 16 and React 19 for the frontend. To handle the backend I implemented a strict dual database architecture: Amazon Aurora: it acts as the ACID compliant system of record for the agent registry, hierarchical permissions, and immutable audit logs. and Amazon Dynamo dB: it serves as the high velocity activity stream, ingesting real time execution telemetry from the agents.

The Challenges I Faced:

The hardest non technical challenge hit right at the finish line is that I completely ran out of my Vercel/v0 credits. Because of this wall, i had to stop the development. While some of the core api gateways, aws integrations, and recursive database queries are fully deployed and real, but might find some features or edge case interactions that feel incomplete or unresponsive.

What I Learned My name is Abhay (Dheeraj), and this project pushed me far beyond standard operations. I learned how to architect dual database systems where relational data and time series events coordinate seamlessly. The amazon aws was completely new for me but somehow managed it. Most importantly, I learned how to ship a live, production-grade product under strict resource constraints.

Built With

Share this project:

Updates