# Virtana lets its AI agents pull in agents from other vendors for IT incidents

published: 2026-10-28T13:05:30.904Z
topic: infrastructure
source: https://siliconangle.com/2026/10/28/virtana-lets-its-ai-agents-pull-in-agents-from-other-vendors-to-work-it-incidents/
url: https://aiagentnews.io/stories/virtana-lets-its-ai-agents-pull-in-agents-from-other-vendors-for-it-incidents

Virtana has added Agentic Observability Enterprise Management, letting its agents invite agents from other vendors into an IT incident investigation. Any agent that supports the Model Context Protocol or Agent2Agent can join, including agents bundled with software a customer already owns.

Virtana, an IT observability company, has added Agentic Observability Enterprise Management to its platform. Its agents can bring other vendors' agents into an investigation when a failure spans several systems. Most enterprise agents ship inside a single vendor's product, so none of them sees the whole outage.

In the company's example, a third-party agent notices a slowed application and uses Agent2Agent to reach Virtana's root cause analysis agent. The diagnosis pointed to a database bottleneck that appeared after a recent deployment. Other specialized agents were then assigned to the fix, and Virtana's software confirmed the application had recovered.

Customers can limit autonomous action to incident types they define in advance, or require engineers to run fixes or approve each agent action. Policy also controls which outside agents may join an investigation and how many tokens they can use, and every action is written to an audit log. Outside agents have been able to query Virtana's monitoring data through its System-Aware MCP Server since February, which covers more than 25,000 real-time metrics.

## Why it matters if you build agents

Developers building incident-response agents can connect to a vendor's root cause agent over MCP or Agent2Agent instead of building that integration themselves. Token limits, per-action approval and audit logging are the controls to review before letting outside agents act. Teams should also confirm which incident types are allowed to run autonomously before turning that on.