Onyx

Onyx builds a security and governance layer that sits across every AI agent and model a company actually uses — sanctioned or not. Its AI Observability module discovers "shadow AI" deployments employees have adopted on their own, with visibility into prompts, responses, and agent actions; AI Security identifies vulnerabilities and applies real-time safeguards; AI Governance lets teams write and enforce compliance policy in plain language; and AI Orchestration tunes agent setup for cost, accuracy, and latency. Its "Guardian Agent" acts as a supervisory AI that continuously finds and remediates risk across deployments. Onyx integrates directly with the major AI platforms — Anthropic, OpenAI, Amazon Bedrock, Microsoft Copilot Studio, Google Gemini Enterprise, Salesforce, and Cursor — and claims to secure over 1.1 million agents across 1.8 million+ covered employees, analyzing 66.2 million+ sessions for threats. It was founded by Maxim Bar Kogan and Gil Elbaz, and raised a $5M Seed and $35M Series A (both March 2026, the A led by Conviction) followed by a $113M Series B in July 2026 led by Bessemer Venture Partners.