The Secure AI Control Plane for the Agentic Era.

Onyx is a secure AI control plane built to close the gap that opened when enterprises began deploying autonomous agents faster than they could see or govern them. As AI shifted from assistive tools to operational agents with real permissions and the ability to act across systems, the risks multiplied: sensitive data exposed through agents, excessive permissions granted by default, and attack surfaces introduced by compromised agent pipelines. Onyx was founded specifically to address this gap, offering a single platform for security and governance teams to discover, oversee, and manage AI agent activity across the enterprise.

The platform covers the full spectrum of enterprise AI deployment — SaaS-based AI tools, cloud-hosted agentic applications, endpoint chatbots and coding agents, and AI components embedded in custom-built software. Onyx provides observability into every prompt, response, and agent action, identifies vulnerabilities and misconfigurations in AI posture, and enforces real-time safeguards as agents operate. Its governance module maps deployments to AI security standards and regulatory requirements and allows teams to define policy controls in natural language. An orchestration layer reduces friction when setting up agents and MCPs, and an ROI measurement capability tracks AI adoption and effectiveness by department.

Onyx's distinguishing position is the control plane framing: rather than securing AI at a single layer or through post-hoc review, it operates as a supervisory layer across all the environments where agents run simultaneously. Its Guardian Agent continuously scans across the platform to identify risks and trigger remediations at scale, which the company positions as a way for large organizations to manage thousands of agents without proportional growth in security headcount. The platform is aimed at enterprise security and governance teams that need end-to-end oversight without slowing down the business units adopting AI.