Metano has launched a security platform for AI agents and raised $12 million in seed funding. Its pitch is to help businesses find agents, understand what they do and block risky actions across systems, rather than securing each corner of the network in isolation.
Watch Desk analysis
What happened
The platform is generally available, SiliconANGLE reports. It combines agent discovery with Hive, a context graph that connects an agent’s tasks, identity, tools, data access and actions. Metano says security teams can use it to assess suspicious behaviour across a whole session, not just flag an isolated action.
A third part applies centrally managed policies and guardrails wherever agents operate. The company says its platform can identify approved and unapproved agents, MCP servers, skills and software packages, and that its detections draw on research from its MARS Labs team. Read SiliconANGLE’s report.
Metano says teams can begin using the platform in under 30 minutes. It also offers a free Eval Edition with 30 days of real-time agent scanning. Those are company claims, not independent performance findings.
Why it matters
AI agents may use legitimate credentials and approved tools while making decisions dynamically. That can make it difficult for security teams to distinguish an authorised task from an agent behaving in an unexpected or risky way. Metano’s approach is to connect identity, context and activity, then give teams a way to enforce controls across the places agents operate.
The launch makes a familiar enterprise-security question more pressing: can organisations let agents do useful work without losing track of what they can access or what they have done? A new control layer may help, but its value depends on what it can actually see and stop in a live environment.
Our read
This is a substantial, AI-specific security launch, not just another promise to put a dashboard around a fast-moving trend. The useful idea is joining discovery, session-level context and enforcement in one platform. Security teams could try the free evaluation to see what agents it finds in their own environment; the tougher test is whether its detections and controls hold up during real work.
Metano is a new entrant, and the performance and deployment claims here come from the company as reported by SiliconANGLE. Buyers should treat the launch as an invitation to test the controls, not proof that agent security has been neatly solved. It rarely stays neat for long.
What to watch
- How accurately Metano finds agents, MCP servers, skills and software packages in customer environments.
- Whether Hive helps teams explain why an action was flagged, rather than simply adding another alert.
- What customers report about enforcement, false alarms and the effort required to deploy the platform.
Discussion spark: Should companies give AI agents broad access and rely on a monitoring layer to catch risky behaviour, or keep agent permissions narrow even if that limits what they can do?
Sources and evidence
- Metano launches with $12M to help enterprises secure AI agents – SiliconANGLE (19 November 2026, 13:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.