As AI agents enter the real world at scale, they will increasingly participate in and shape our economic systems.
We operate as a small systems lab. Our team consists of engineers and researchers who have designed and studied AI agents across diverse domains.
At Dunia Labs, we start from a simple premise: meaningful autonomy requires structure. We build the infrastructure that lets AI agents operate reliably — not by constraining what they can do, but by giving them a foundation to do it safely.
Peta is infrastructure for production AI: an operations and permissions layer that developers can integrate directly into their existing stacks.
peta.ioGaur is a terminal supervisor with GPU-accelerated rendering, autopilot, and remote control.
gaur.devValva is a credential plane for AI agents: server-side secret injection, policy enforcement, and audit for every tool call.
valva.ioOctvex is an operations layer for AI agent work: task orchestration, approvals, and end-to-end audit. Self-hosted today, with a managed option ahead.
octvex.com