Virtana Enables AI Agents to Collaborate Across Vendor Boundaries in IT Incident Response

Virtana, an observability platform focused on information technology, unveiled Agentic Observability Enterprise Management, a feature that permits its artificial intelligence agents to recruit agents from rival vendors when responding to IT incidents. The announcement targets a structural weakness in how organizations have rolled out AI agents to date.

Enterprise adoption of AI agents has typically followed a single-vendor model, with agents embedded directly into a particular product—such as an IT service management platform. This approach creates blind spots: when an outage affects infrastructure from multiple vendors, each vendor's agent only perceives part of the problem, leaving engineers to manually piece together the full picture.

Virtana's new capability allows one of its agents that detects an issue to bring in agents from other vendors to collaborate on the investigation. Any agent supporting the Model Context Protocol or the open Agent2Agent protocol can participate, including agents that already ship with software companies have deployed.

Since February, external agents have had the ability to access Virtana's monitoring information through its System-Aware MCP Server, which delivers more than 25,000 real-time metrics. Agents brought into an incident also receive a unified perspective on application dependencies and underlying infrastructure through the System Dependency Graph, which Virtana introduced alongside the MCP Server. The company expanded its monitoring capabilities to cover AI workloads a month later.

In Virtana's scenario, a third-party agent detects application slowness and engages Virtana's root cause analysis agent—located via Agent2Agent—to diagnose the problem. The investigation identifies a database constraint that emerged following a recent code release. The root cause agent can field follow-up questions before additional specialized agents are enlisted to implement repairs. Virtana's platform then validates that the application has returned to normal operation.

A traditional IT incident response war room requires hours to assemble and operate, according to Chief Product Officer Amitkumar Rathi. He stated that a coordinated team of agents can accomplish the same resolution "in milliseconds."

This level of autonomous action applies only to incident categories that customers have predefined. Organizations preferring greater oversight can mandate that engineers execute the fixes agents recommend, or require approval before each agent action executes. Configuration controls also govern which external agents participate in investigations and the token budget each agent receives. All agent actions are recorded in an audit trail.

Virtana, formerly known as Virtual Instruments before rebranding in 2019, began as a venture-backed provider of IT infrastructure performance monitoring tools. The company attracted $73 million in investment during 2022, with backing from Atalaya Capital Management LP and Benhamou Global Ventures, among others.