Onyx Security: A Secure Control Plane for the Agentic Era

A new, ungoverned attack surface

Enterprises are racing to put AI agents to work across their operations: writing code, answering customers, moving data, and increasingly taking actions on their own. The productivity unlock is seemingly enormous. So is a category of risk that existing security tools were never designed to handle. Agents operate autonomously, reach across cloud, endpoint, and SaaS, and frequently run with far more access than the task in front of them requires. The result is a fast-expanding attack surface that most organizations can neither see nor control. According to third-party studies, the majority of enterprises are already exposing sensitive data through AI agents, and the overwhelming majority run agents with excessive permissions. Traditional security and governance stacks were built for humans and static applications. They can break down when the actor is a non-deterministic model that can be prompted, manipulated, or simply make a mistake at machine speed. We believe securing autonomous AI will be one of the defining infrastructure challenges of the coming decade and that meeting it requires a platform built for the problem from first principles.

A secure control plane for AI

Founded in 2024 and headquartered in Tel Aviv, Onyx has built a secure control plane that sits across an enterprise's entire AI footprint to discover, secure, govern, and remediate AI agents and models at runtime. Rather than bolting policies onto a perimeter, Onyx uses proprietary, purpose-built models to understand what agents are actually doing, intervene in real time when behavior crosses a line, and give security teams a single place to enforce policy across every prompt, response, and action.

The platform spans the full lifecycle of enterprise AI: discovering sanctioned and shadow AI, hardening it against threats and misuse, enforcing governance and regulatory requirements through controls defined in plain language, orchestrating agents and model context for cost and performance, and measuring adoption and ROI. At its center is the Onyx Guardian Agent, a supervisory layer that continuously hunts for risk and remediates issues so organizations can manage thousands of agents at scale without slowing the business down. The vision is simple but powerful: let enterprises double down on AI adoption without losing control of it.

Built by AI Veterans

The most disruptive security companies often are not built by security veterans alone. Securing autonomous AI is, at its core, an AI problem, and solving it requires people who understand how these systems actually think, fail, and can be steered. That is the new set of eyes we found at Onyx. Maxim Bar Kogan (Co-Founder and CEO) served in an elite military technology unit before scaling product and engineering at Mixtiles and serving as CTO of Segmanta. Gil Elbaz (Co-Founder and Chief AI Officer) has spent over a decade at the frontier of Machine Learning, including as an AI Architect at NVIDIA and as CTO of Datagen. We believe their pairing of hard-won security instincts with deep, first-hand experience building advanced AI is exactly what this problem demands.

Partnering with Onyx

We believe the rise of autonomous AI is the most significant platform shift since the move to the cloud and that the companies enabling enterprises to adopt it safely will be foundational to the modern technology stack. As software is increasingly written, and now run, by machines, trust and governance stop being back-office concerns and become mission-critical infrastructure.

TCV is proud to have invested in Onyx's Series B, led by Bessemer Venture Partners, alongside existing investors Cyberstarts and Conviction. At TCV, we gravitate toward founders who are obsessed with product and unafraid of hard problems. We believe Maxim and Gil pair that obsession with a rare clarity about the category they are defining, and we look forward to supporting the entire Onyx team on the journey ahead.

The views and opinions expressed are those of the author and do not necessarily reflect those of TCMI, Inc. or its affiliates ("TCV"). Such statements have not been independently verified and are subject to change. This blog post contains forward-looking statements, which are based on current expectations, forecasts and assumptions that involve risks and uncertainties that could cause actual outcomes and results to differ materially. TCV disclaims any obligation to update or revise any forward-looking statement to reflect changes. This blog post is not an offer to sell or the solicitation of an offer to purchase an interest in any private fund managed or sponsored by TCV or any of the securities of any company discussed. The investment described above may not be representative of all investments made by the TCV Funds and is not necessarily representative of any future investments that may be made by the TCV Funds. No assumption should be made that the investment identified was or will be profitable. For a complete list of TCV investments, please visit www.tcv.com/all-companies. For additional important disclaimers regarding this blog post, please see "Informational Purposes Only" in the Terms of Use for TCV's website, available at www.tcv.com/terms-of-use.