All community stories

Tech

A security layer between AI agents and company databases

Issak Raj builds software through GlitchGo, and his main project is AdminZero, a zero-trust security gateway that inspects what AI agents are allowed to do before their requests reach production systems.

Ravi Issak Raj, Kakinada, India

Companies are giving AI agents access to databases, APIs and internal systems. An agent is useful when it does the task correctly. An unsafe or manipulated request is another matter: it can expose sensitive data, run destructive database operations, or do something nobody intended.

AdminZero is Issak Raj's answer, a security layer that sits between AI agents and company systems. It inspects and controls database operations before they reach production, using deterministic analysis such as SQL AST inspection, alongside PII protection and audit logging. The point is to give a company control and visibility over what its AI agents actually did, rather than trusting a model with direct access to critical infrastructure.

“I'd rather keep building real things and learn from what happens when they meet real users.”

Ravi Issak Raj

He is a second-year B.Tech Computer Science Engineering student, and his main initiative is GlitchGo, through which he takes on software projects, websites, deployment and technical work. Building it has taught him what it takes to move an idea to something in front of users, and what goes wrong in between.

He writes as well as builds, making technical content on AI, software and cybersecurity, mostly as a way of working out how to explain what he learns from his own projects.

And he has published two books. The first, SQLite + Next.js Production Blueprints, is an ebook on the practical and production use of both, and has taken 11 orders so far. The second, Ship It With AI, launched recently as a paperback and is about building and shipping software with AI, which is where he thinks the craft is moving.

Around those sit a number of smaller experiments across software, AI, automation, student productivity and cybersecurity, which he uses to find out how ideas become products.

All of it points one way. He wants to be an entrepreneur and build technology products and companies of his own, and he is accumulating the experience for it while doing the degree, by building, publishing and putting the work in front of real people.