New Relic has launched a suite of observability integrations for GitHub Copilot, including a public preview of Change Tracking for GitHub Copilot.
The new tools bring telemetry into developers' workflows from coding through deployment, aiming to link software changes with production performance and incident response inside GitHub rather than in a separate dashboard.
A central part of the launch is Change Tracking for GitHub Copilot, which connects production telemetry with deployment activity. It can run pre-merge checks against a baseline, apply a canary gate after deployment and detect regressions against real-time performance data.
If regressions are found, the system can route them to GitHub Issues and Copilot for fixes, and send updates to Slack. The approach is intended to reduce manual triage and shorten review cycles when engineering teams investigate performance issues after software releases.
The announcement comes as software groups weigh the benefits of AI coding tools against the operational risks that can emerge after code reaches production. New Relic cited its 2026 State of AI Coding research, which found that 94% of leaders rated AI-generated code as higher quality at review time, while 78% reported more incidents after deployment.
That gap between review and production behaviour is one of the issues New Relic is trying to address by tightening links between coding agents, deployment activity and live system telemetry. Rather than relying only on code review, engineers are being given data on how a release affects downstream services once it is running.
New Relic also introduced an integration between its Autopilot tool and GitHub MCP Server. Built on the Model Context Protocol standard, the integration allows Autopilot to use codebase context while triaging alerts, identifying root causes and recommending fixes.
The integration is intended to give developers more direct guidance during incident response. By combining operational data with repository context, it aims to keep engineers within their existing workflow while they investigate problems.
Another part of the release is Preflight for GitHub Copilot, which New Relic described as an open-source tool for monitoring how developers use coding agents during the coding process. It provides visibility into model efficiency, latency and token consumption in a local integrated development environment and command-line interface.
According to New Relic, the tool gives teams a unified local view of coding agent usage so they can assess efficiency, spot errors during execution and compare suggestion accuracy across different large language models. The data can also be used to set alerts and manage costs.
Shift in workflow
The broader theme of the launch is a move to place observability data directly inside development tools rather than keeping it in a standalone monitoring product. That reflects a wider shift in software development, where AI-assisted coding is increasing the volume and speed of changes while putting more pressure on teams to understand production effects earlier.
Camden Swita, Head of AI at New Relic, said the company sees observability as part of the development process rather than a separate operational layer.
"As AI accelerates development, engineers and agents need a shared understanding of how their software behaves in production," said Camden Swita, Head of AI at New Relic. "While traditional observability locks critical data inside closed SaaS dashboards, headless New Relic brings trusted operational intelligence directly into the tools and workflows where software is built and delivered. The new GitHub integrations put that strategy into practice, helping teams understand the production impact of any deployment. Our vision is for observability to inform development decisions as they happen, giving engineers and AI agents the context they need to turn faster coding into reliable software delivery."
Change Tracking for GitHub Copilot is available in public preview, while Autopilot for GitHub MCP Server and Preflight for GitHub Copilot are generally available.