New Relic announced Autopilot, an autonomous SRE agent that automatically triages production incidents, identifies root causes, and scopes possible remediations. The agent begins investigating the moment an alert fires, according to the New Relic Now June 2026 announcement, compressing investigations that typically take hours into minutes.
How Autopilot Works
Autopilot is built on New Relic’s observability data substrate and ships with domain-specialist agents for Kubernetes troubleshooting, Kafka diagnostics, and cross-stack root-cause analysis, with more planned. Each investigation produces a structured account grounded in concrete telemetry data, not model speculation.
The agent integrates external context through Model Context Protocol (MCP) connections to Jira and GitHub, pulling in code changes and deployment details to narrow down problems. It also maintains long-term memory, capturing tribal knowledge from past incidents and runbooks so that institutional context persists across team members and shifts.
“Operations are going headless. AI agents won’t log in to view dashboards. They’ll pull what they need through APIs, reason about it, and act,” Camden Swita, Head of AI at New Relic, told Digital Terminal. “For teams who’d rather we run the agent, there’s New Relic Autopilot, our expert agent operated for you.”
Ground Truth: Letting External Agents Access New Relic Data
Alongside Autopilot, New Relic introduced Ground Truth, a suite of agent-optimized tools that gives external agents direct access to New Relic’s observability insights. GitHub Copilot, Claude Code, AWS DevOps agents, and custom orchestrators can all connect through Ground Truth to pull telemetry data without going through dashboards or basic query tools.
New Relic claims Ground Truth is designed for token efficiency, helping agents discover insights with fewer tool calls. One large enterprise customer self-measured a 1.1% error rate across 1,300+ users, according to the Digital Terminal report.
The Vendor Bet on Autonomous Incident Response
The launch reflects a broader vendor bet: that production-grade autonomous agents for high-stakes operations environments are ready for enterprise deployment. New Relic’s own data supports the urgency. The company’s 2026 State of AI Coding report found that 78% of technology leaders report an increase in production incidents after deploying AI-generated code, even though 93.5% rate that code as higher quality during review. The gap between review-stage confidence and production-stage reliability is exactly the problem autonomous incident response is designed to close.
Autopilot runs inside the New Relic platform, in Slack, and via automated workflow actions triggered by severity thresholds, deployments, or schedules. Both Autopilot and Ground Truth are listed as available summer 2026.