Apa yang terjadi
On Nov 19, 2026 Metano launched with $12M seed (Sierra Ventures-led) and a generally available platform for securing AI agents across endpoints, cloud, and SaaS. It unifies agent discovery (approved and unapproved agents, MCP servers, skills, packages, privileges and exposure), a real-time context graph called Hive that correlates tasks, identities, tools, data access and actions across a session to judge whether behavior matches intent, and centrally managed enforcement/guardrails that operate where agents run (including activity that bypasses network gateways). Controls can be generated from a developer-specified agent spec or baselined from ~a week of observed behavior, start in audit mode, and full inline enforcement for homegrown agents requires one line of its SDK. Detections map to MITRE ATLAS and OWASP Top 10 for Agentic Apps; platform is SOC 2 Type 2 and ISO 27001 certified.
Mengapa penting
A new entrant competing on session-level intent rather than single-action checks, explicitly targeting the gap where agents act with legitimate credentials and approved tools but drift from purpose. It provides a multi-surface (endpoint/cloud/SaaS) control point and a practical audit-mode-first onboarding path.
Cakupan penerapan
Security teams standing up agent governance. Evaluate the free 30-day Eval Edition for agent/MCP inventory and behavior baselining before committing, especially where homegrown agents require SDK integration for full inline enforcement.