Sometimes, you just can’t get a question out of your head. For me, that question came after ChatGPT launched and I realized the extraordinary impacts AI agents would have on society. I realized that these systems were going to become smarter than us and that eventually, we would be relying on them for things that really matter, even managing our water, food and electricity supply. AI agents will be making decisions that will be harder and harder for us to understand, and that will affect every aspect of our lives.
And I couldn’t stop thinking: “When something smarter than you is responsible for the things you depend on, how do you stay in control?”
That question became my obsession—and eventually, was the inspiration for starting Onyx and partnering with my co-founder and our Chief AI Officer Gil.
After a year and a half of quietly building, today we’re stepping out of stealth with $40 million in funding from Conviction Partners and Cyberstarts, a team of 70, and an AI control plane that’s already being used by Fortune 500 companies to govern their AI agents.
AI Agents: A New Class of Actor in the Enterprise
In 2026 every company, whether they planned to or not, is becoming an agent operator. AI agents are now within reach of every employee. They are writing code, triaging IT tickets, managing customer workflows, and making decisions across platforms, tools, and identity boundaries at machine speed. They have access to the most sensitive data sources and systems, and are operating at increasing autonomy and scale. And unlike traditional software, they don’t behave the same way twice. Their actions are shaped by the non-deterministic LLMs that power them, dynamic context, untrusted inputs, and tool outputs that can’t be predicted in advance.
An entirely new layer of operational infrastructure, ungoverned. The security tooling that exists today was built for a world of deterministic software and human-driven workflows. It cannot see this new surface, let alone control it. This gap is growing every week, and new solutions are needed.
What We’ve Built at Onyx
Onyx provides a single, unified enforcement layer that continuously discovers every AI asset across the enterprise—agents, models, AI-powered applications, MCP-connected tool ecosystems—and turns governance policies into enforceable controls that operate at runtime.
The Onyx Guardian Agent is at the heart of the platform. It oversees all other AI agents in the customer’s environment, and can block unsafe actions, require human approval, narrow an agent’s scope, or redirect it toward a safer path—all before anything reaches a downstream system.
With our solutions in place, teams can move fast and adopt AI aggressively because they have the infrastructure in place to do it safely.
Our Team
Our proudest achievement is the incredible team of over 70 employees we’ve assembled to tackle this challenge who are a mix of cybersecurity and AI experts.
I spent my career at the intersection of AI and offensive cyber, mostly in Unit 8200 of the IDF. Gil is an AI architect who built systems at the frontier of modern AI, most recently leading agentic AI work with the CTO of Nvidia. We connected running together and having long philosophical discussions that all circled back to the same point: agents are going to become critical infrastructure, and someone needs to build the control layer to make them safe.
That mixture of deep security DNA paired with frontier AI experience is the foundation of our company. It’s reflected in our team, where AI researchers training models sit alongside security researchers.
What’s Next
At Onyx, we believe that we’re headed toward a future where every employee will have an army of AI agents working alongside them. A future where small teams can compete with large organizations because of how effectively they orchestrate AI. That future is exciting, but it could also be dangerous if we get the control problem wrong.
Onyx exists to make sure we get it right. If you’re deploying agents in production and thinking about what governance looks like, we’d love to show you what we’ve built.
Get in touch to book a demo at onyx.security/get-a-demo