The AI era isn’t just changing software—it’s complicating it. As systems grow more autonomous and interconnected, traditional monitoring tools are buckling under the weight of “agentic AI” architectures—where multiple AI agents work together, pulling data, invoking tools, and reacting in real time.
To meet that challenge, New Relic has launched two major innovations—Agentic AI Monitoring and the AI Model Context Protocol (MCP) Server—at the intersection of observability and AI. Together, they aim to help developers understand what’s actually happening under the hood of their agent-driven systems, instead of guessing where things went wrong.
Turning Complexity Into Clarity
Agentic AI systems—used in everything from autonomous coding to customer service bots—are inherently messy. One agent’s output becomes another’s input; one hallucination can ripple through the chain before anyone notices. According to New Relic’s 2025 Observability Forecast, 54% of organizations are now using AI monitoring tools (up from 42% in 2024), but most still struggle to visualize or control multi-agent interactions.
New Relic’s Agentic AI Monitoring addresses that blind spot head-on. It offers a unified view of every agent, tool call, and dependency within an organization’s AI mesh—showing how agents communicate, which tools they invoke, and where latency or failures occur. A visual Agents Service Map helps teams trace issues across multiple MCP servers and contexts, cutting through the noise of distributed decision-making.
Unlike standard LLM monitoring, which stops at the model level, this new capability extends to the infrastructure and services beneath the AI, providing full-stack observability that bridges the gap between code, data, and intelligent agents. In practice, that means DevOps and AI engineers can pinpoint issues faster, improve uptime, and validate ROI on AI deployments.
A Common Language for AI Agents
The second announcement—the New Relic AI MCP Server—introduces something even more transformative: a standardized way for AI assistants like ChatGPT, Claude, GitHub Copilot, and Cursor to directly access New Relic’s observability data.
That integration gives these tools real-time visibility into system performance—turning conversational assistants into fully informed copilots that understand what’s happening in production. Instead of engineers toggling between dashboards, they can ask questions like, “Why did latency spike in our payments API?” and get actionable, observability-backed answers from within their coding environment.
It’s a move that effectively democratizes observability, bringing it out of the ops silo and into the flow of everyday engineering work. As IDC’s Stephen Elliot notes, “By creating an intelligent feedback loop where AI systems become more observable and reliable—while observability platforms become more intelligent—the industry can innovate with confidence.”
Observability in the AI Age
The broader context is urgent. High-impact outages now cost an average of $2 million per hour, or roughly $33,000 per minute, according to New Relic’s report. With AI systems increasingly running mission-critical workloads, the margin for silent failures is vanishing.
Brian Emerson, New Relic’s Chief Product Officer, summed it up bluntly: “The convergence of AI workloads, cloud-native architectures, and real-time data processing has created a perfect storm of complexity. Our latest innovations empower enterprises to adopt AI systems that create business value—without cutting into the bottom line.”
Going Beyond Anomalies
In addition to its agentic tools, New Relic also rolled out Outlier Detection, a companion to its anomaly detection suite. The feature goes beyond simply flagging deviations—it analyzes abnormal behaviors and surfaces the root causes faster, letting teams prioritize incidents before they hit users. It’s another step toward the company’s goal of “intelligent observability”—where systems not only watch but learn.
Why It Matters
As enterprises move from single-model AI integrations to networks of autonomous agents, the monitoring problem shifts from “what’s wrong with the model?” to “how do these systems behave as a collective?” With Agentic AI Monitoring and the MCP Server, New Relic is positioning itself as the first major observability platform designed for that reality.
By making AI more transparent to humans—and observability tools more accessible to AI—New Relic is effectively giving modern software a nervous system fit for the agentic age.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
