About
Janus Labs is an independent Australian research and software venture focused on AI agent reliability, evaluation, and technical writing.
Alex Perry is a product and technology leader based in Sydney. Janus Labs is his independent research and software venture focused on AI agent reliability, evaluation, and technical writing.
His background spans customer operations, product delivery, privacy and controls, and large-scale systems change. That mix is what shaped the central Janus question: how do you add governance to a cognitive system without degrading the cognition?
The Janus Protocol began as a practical tool for stopping AI agents from looping on the same failed approach. It became a research program once the patterns repeated often enough to need clearer language, stronger classification, and better measurement.
Australian proprietary limited company.
Sydney, Australia.
2026. Research program active since late 2025.
Janus Labs publishes open research, maintains the Janus Protocol, and builds evaluation infrastructure for teams working with AI agents.
The work sits at the intersection of research, protocol design, and practical evaluation. The name comes from the Roman god of transitions and duality: two faces looking in opposite directions at once. It suits a system built on the idea that generation and critique should stay separate.
The goal is higher verifiability, not more ceremony. If every safeguard adds constant friction, the system may look controlled while getting worse.
The field often treats anecdotes, observations, principles, and validated results as if they carry the same weight. Our taxonomy is designed to keep those distinctions visible.
Reliable review is easier when the critique function is architecturally distinct from the generation function. That is why Janus treats the Builder and Watcher as different jobs.
Janus Labs is open to research collaborations, technical writing opportunities, and conversations with teams evaluating AI agent reliability.