New Relic used Microsoft's annual Build developer conference in San Francisco to announce a sweeping set of integrations with Azure and GitHub, extending its observability platform deep into the AI agent lifecycle and reinforcing a 14-year strategic alliance with Microsoft. The company also reported strong double-digit year-over-year growth in committed cloud revenue, signaling robust enterprise demand for tooling that can tame the complexity of AI-powered operations.

The announcements, made during a series of sessions and demos at Build 2026, position New Relic as a first-class observability provider for the emerging paradigm of agentic AI—autonomous software entities that plan, reason, and take action. By embedding monitoring, tracing, and intelligence directly into Azure’s AI services and GitHub Copilot workflows, New Relic aims to give developers and operations teams the same depth of insight into AI agents that they’ve long had for traditional applications.

A 14-Year Partnership Goes Agentic

New Relic’s relationship with Microsoft started with .NET and Windows Server monitoring in the early 2010s. Over the years, it expanded to cover every major Azure service, from virtual machines and containers to serverless functions and SQL databases. Today, New Relic is a native part of the Azure portal, and thousands of joint customers rely on the integration for full-stack observability.

At Build 2026, the partnership took a decisive turn toward AI. “For over a decade, we’ve worked side by side with Microsoft to deliver world-class observability for Azure workloads,” said New Relic CEO Bill Staples during the event’s opening keynote. “Now, with the explosion of AI agents and copilots, the definition of ‘workload’ has changed. Our mission is to bring the same level of visibility, reliability, and business insight to every AI-powered process, from prompt to action.”

The centerpiece of the announcement is a new service tier called New Relic AIM for Agentic Systems. Available immediately in public preview for Azure customers, it delivers purpose-built instrumentation for AI agents built on Azure AI Foundry, Azure OpenAI Service, and custom models running on GPU clusters. It also introduces native integration with GitHub Copilot Extensions, closing the feedback loop between code generation and runtime performance.

Deep Azure Integration: Observability as a First-Class AI Service

New Relic AIM for Agentic Systems plugs directly into the Azure AI stack without requiring developers to add custom telemetry code. It auto-instruments Azure AI Foundry’s agent orchestration engine, capturing every step in an agent’s decision tree: prompt templates, retrieval-augmented generation (RAG) calls, tool usage, memory accesses, and final outputs.

Key capabilities of the Azure integration include:

  • Agent Traces: A new trace type that models an agent’s entire execution as a directed acyclic graph, making it easy to spot loops, hallucinations, and costly chains of reasoning. Traces link back to specific Azure OpenAI deployments and token usage, giving FinOps teams precise cost attribution.
  • Safety and Alignment Monitoring: Built-in checks for content safety filters, toxicity scores, and jailbreak attempt detection, all fed into New Relic’s dashboard. Alerts can be configured to pause agent deployments that exceed enterprise risk thresholds.
  • Azure Monitor Bridge: Bidirectional telemetry flow between New Relic and Azure Monitor. Azure Monitor signals—including GPU utilization, model latency, and prompt caching hit rates—are unified under New Relic’s entity model, while New Relic’s application-level insights are exported back to Azure for use in Azure Policy evaluations and auto-remediation runbooks.
  • Copilot for Azure Integration: New Relic is one of the first observability partners to join the Copilot for Azure ecosystem. IT operators can ask natural-language questions like “Show me the token efficiency of my finance agent over the last 24 hours” directly in the Azure portal and get curated New Relic charts, logs, and recommendations without leaving the console.

During a live demonstration, a New Relic engineer showed how a supply-chain agent powered by Azure OpenAI Service began producing increasingly verbose responses as its memory grew. The New Relic AIM dashboard immediately flagged the trend, highlighted a 37% increase in token consumption per task, and recommended truncation of old context—a fix that was applied automatically via a low-code automation template.

GitHub Copilot Extension: Closing the Code-to-Runtime Loop

Perhaps the most forward-looking announcement is the New Relic extension for GitHub Copilot, which embeds runtime performance data directly into the developer workflow. Available as a Copilot expansion package, the extension brings three core experiences:

  1. Code-Level Performance Insights in Pull Requests: When a developer opens a pull request that modifies a service instrumented by New Relic, the extension annotates the diff with performance impact predictions. For example, a line that adds a new database query might be flagged with “Estimated impact: +12 p99 latency under load. Consider adding a read-replica cache.” The predictions are generated by a fine-tuned model trained on billions of production traces.
  2. Natural Language Observability Queries: Developers can ask Copilot Chat questions like “What’s the error rate for the checkout service in production since last deploy?” and receive New Relic-sourced answers, complete with links to drill-down views. The extension handles authentication context, so users see only data for which they have New Relic permissions.
  3. AI-Generated Runbook Automation: When Copilot identifies a routine pattern (e.g., a memory leak or a flaky API), it can propose a remediation runbook written in Pulumi or Bicep, which New Relic can then execute via its applied intelligence actions. This creates a self-healing loop where code issues detected in production can trigger patches that are proposed back in GitHub, ready for review.

The Copilot extension is designed to help platform teams mature their AI Ops capabilities while keeping the human in the loop. “We’re not trying to replace SREs,” Staples noted. “We’re giving them superpowers. The goal is to make the feedback from production so immediate and actionable that fixing issues becomes a trivial part of the daily flow.”

Observability for AI Agents: Beyond Traditional APM

Agentic systems break the traditional request-response model that APM tools were built around. An AI agent might invoke dozens of external tools, loop through multiple reasoning cycles, and run for minutes or hours before completing a task. Standard metrics like throughput and error rate become meaningless; what matters is the quality, efficiency, and business outcome of each agent invocation.

New Relic’s agentic observability approach tackles this with three innovations:

  • Outcome-Based Service Level Objectives (SLOs): Instead of measuring response time or uptime, teams define SLOs for agent tasks, such as “95% of customer support agent sessions must end with issue resolution within three conversation turns.” New Relic tracks these outcomes and burns error budgets when agent performance degrades.
  • Cost-Effectiveness Dashboards: A dedicated FinOps view for AI shows token consumption per agent, per department, and per business unit, with recommendations to switch to cheaper models, optimize prompt length, or cache common queries. One early adopter reported a 22% reduction in monthly OpenAI spend after implementing New Relic’s suggestions.
  • Agentic Live Archives: Every agent execution is recorded in a queryable format that retains full context—prompts, intermediate steps, tool calls, and outputs—for compliance and debugging. The archives can be searched by natural language, e.g., “Find all sessions where the customer mentioned a refund and the agent quoted the expired policy.”

These capabilities rely on New Relic’s existing time-series database and stream processor, which have been retrofitted to handle the high-cardinality, variable-length telemetry that agents produce. The company disclosed that it processes over 2 billion agentic spans per day across its customer base, a number that has tripled in Q4 alone.

Growth and Strategic Rationale

New Relic’s Build presence coincides with a period of renewed commercial momentum. The company reported double-digit year-over-year growth in committed cloud revenue, driven largely by enterprises migrating AI workloads to Azure and standardizing on a single observability platform for both traditional and agentic applications. While exact figures weren’t disclosed on stage, industry analysts estimate New Relic’s monthly recurring revenue from its AIM product line has crossed $100 million.

The partnership structure has also evolved. Microsoft is now a “go-to-market acceleration partner” for New Relic’s agentic products, meaning Azure sales teams will actively position New Relic AIM alongside Azure AI Foundry deals. In return, New Relic will deepen its commitment to Microsoft’s platform, including adoption of Azure as its primary AI training infrastructure and increased usage of GitHub for internal development.

For Windows and .NET developers—the core audience of windowsnews.ai—the announcements are particularly significant. New Relic already offers industry-leading instrumentation for .NET 9 and modern Windows Server 2025 workloads. The new agentic capabilities will be available as NuGet packages that plug directly into ASP.NET Core web applications hosting any part of an AI agent pipeline. A workshop at Build, “Agentic .NET: From Code to Intelligent Ops,” walked 300 attendees through the end-to-end experience.

Competitive Landscape and What Comes Next

The observability market is fiercely competitive, with incumbents like Datadog, Dynatrace, and Splunk all racing to add AI monitoring features. But New Relic’s first-mover advantage in AI agents—coupled with its Microsoft-exclusive integrations—could carve out a durable niche.

At Build, New Relic committed to making the agentic observability capabilities generally available on Azure by October 2026, with support for other cloud AI platforms (AWS Bedrock, Google Vertex AI) planned for early 2027. The GitHub Copilot extension will remain in preview until it can ingest data from non-New Relic backends, a requirement that Microsoft is said to be heavily influencing.

For IT teams navigating the shift from assisting humans with software to managing fleets of autonomous agents, Build 2026 made one thing clear: observability is no longer just about “is my server up?” It’s about asking, “Is my AI doing what I expect, at the right cost, and without unintended consequences?” New Relic’s latest moves give practitioners the tools to answer that question with confidence—and to fix what’s broken before the customer ever notices.

Developers interested in trying the preview can sign up via the New Relic Azure Marketplace listing or install the GitHub Copilot extension from the Visual Studio Code marketplace. Full documentation and quickstart guides for agentic observability are available on New Relic’s developer portal.