New Relic Autopilot is opening a new front in observability: letting AI agents—not just humans—handle end-to-end incident management by working directly on platform data, not dashboards. This is a break from the classic model, where response speed and accuracy live or die on whether an engineer saw the alert, traced the root cause, and flipped fast between metrics and logs. For DevOps and SRE teams pushing into AI-first operations, New Relic’s June 2026 launch of Autopilot, coupled with Ground Truth, is the first real sign of the dashboardless future. AI can work incidents faster, with stronger context, and feed its learnings back into the substrate—raising the bar for uptime and resilience. SRE teams working at scale, or those facing noisy on-call handoffs and sprawling architectures, have real reason to pay attention.
What is New Relic Autopilot and how does it use AI for incident management?
New Relic Autopilot is an AI-powered site reliability engineering (SRE) agent that takes over core incident response tasks. Announced in June 2026, Autopilot is built to analyze alerts as soon as they fire, triage incoming incidents, identify root causes, and scope out remediation—all without human-in-the-loop dashboard review.
The actual workflow is agentic: when a monitored system hits a threshold or error and an alert triggers, Autopilot hooks into the event. From there, it pulls the relevant observability data—metrics, logs, traces—using New Relic’s standardized API substrate. Autopilot incorporates specialized SRE logic for Kubernetes and Kafka and is designed for cross-stack root-cause analysis, with support for more platform domains on the roadmap.
A critical feature is how it adapts knowledge: Autopilot uses New Relic Knowledge to ground its actions in an organization's runbooks and retrospectives. It extends context out to Jira and GitHub—ingesting both code- and ticket-driven context via Model Context Protocol. It holds long-term memory for operational learnings that would otherwise get siloed per engineer or team. For human responders, this compresses the “what broke?” and “how do we fix it?” loops.
The upshot: Autopilot lets teams hand off rote incident analysis at machine speed, raising the floor for response time and enabling engineers to focus on higher-use tasks. For the first time, incident context, reasoning, and remediation paths live in the substrate—not just in someone's notebook or dashboard history.
How do AI observability agents like Autopilot replace traditional dashboards?
AI agents, including New Relic Autopilot, skip the dashboard UI entirely. Instead of having engineers log in, scan charts, and click through breadcrumbs, these agents tap into observability platforms through pure APIs. The pattern: headless operation with no need for human interface.
Camden Swita, New Relic’s Head of AI, makes the shift explicit: “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.” The dashboard—once the centerpiece of NOC workflows—is now the fallback, not the default.
What does this enable? First, it untangles speed and alert fatigue. AI agents process incoming signals at wire speed, reason across sources, and launch investigation and triage steps in seconds, not minutes. Second, it makes automation precise: incident context is constructed from structured telemetry, not layers of undocumented analyst steps. Third, it enables full auditability—because reasoning and actions are captured in the substrate, routed through APIs, and queryable after the fact.
For SRE and DevOps, this changes the calculus. Instead of configuring dashboards for every edge case, the substance shifts to enforcing data cleanliness, permissioning API access, and governing the surface between human and agent action. Dashboards will still exist, but as one view among many. The new center of gravity is the headless, API-first SRE agent loop.
What role does New Relic Ground Truth play in enhancing observability for AI?
New Relic Ground Truth is the foundation for AI observability. Where Autopilot handles actions and response, Ground Truth standardizes and governs the telemetry data that those agents consume. Without a clean substrate, all the automation in the world is brittle.
In practice, Ground Truth creates a vetted, unified layer of observability data that serves as the single source of truth for both human and AI consumers. This means that alert signals, metrics, logs, and traces follow shared semantics and governance, reducing noise and accidental drift in incident reasoning.
For AI-driven agents, the payoff is immediate: grounded context. Autopilot and its peers don’t have to guess at meaning or wrangle partial data—they receive telemetry that is scrubbed, labeled, and versioned for consumption. This makes AI incident management reliable, lowering the risk of hallucinated root causes or missed dependencies.
The real value: incident automation becomes trustworthy. When the underlying telemetry matches the operational state, AI agents’ conclusions can be audited and traced. Ground Truth is the backbone for aligning automatable action with what’s actually happening in production.
How has New Relic simplified OpenTelemetry adoption for AI observability?
OpenTelemetry is now the lingua franca for observability—especially in mixed cloud and on-prem architectures. New Relic’s recent platform push gives enterprises a less-disruptive way to migrate observability workloads from bespoke or vendor-locked sources to standardized OpenTelemetry.
The pain point: Enterprises wanted open telemetry for portability, but rewiring global monitoring around new shims and data contracts risked operational drag. New Relic’s response was to make the path less painful—abstracting integrations and supporting hybrid ingestion, so teams don’t have to burn down existing pipelines overnight.
For AI observability, and agent-driven SRE, this enables two things. First: confidence that your full telemetry substrate (past and present) can support automation, because data is cleaned and mapped to OpenTelemetry standards. Second: AI agents like Autopilot can reliably access telemetry without needing custom scrapers or retrofitted dashboard parses. Mixed-mode observability—where legacy, SaaS, and AI agents all talk the same language—becomes operationally real.
Adopting OpenTelemetry the New Relic way means less migration drama, with the operational upside of unified data for both human and agent-driven incident response.
How can DevOps and SRE teams use New Relic Autopilot today?
Getting Autopilot into the workflow is about wiring automated incident response into the team’s core operations loop. Here’s the high-order path:
- Enable Autopilot in your New Relic environment: Autopilot is an add-on to the New Relic platform, so activation is surfaced in the admin console or via API. Turn it on for critical services first.
- Connect alert policies: Route high-signal, actionable alerts into Autopilot. This is about signal quality: only the right classes of incidents should trigger autonomous triage.
- Integrate with runbooks and retrospectives: Plug operational knowledge into the system so AI actions reference real, validated playbooks.
- Configure external context fetches: Link Jira, GitHub, and other Model Context Protocol sources to feed incident data and work-tracking context.
- Automation in practice: Once enabled, Autopilot picks up eligible incidents as they hit, pulls in the relevant telemetry, clusters similar events, attaches context, and drafts remediation steps—either serving up recommendations or auto-executing (depending on policy).
Example minimal setup (pseudocode):
# 1. Enable Autopilot via New Relic CLI or dashboard
newrelic autopilot enable --services critical-app-1, critical-app-2
# 2. Route alert policies
newrelic autopilot alert-policy attach --policy "High CPU Usage" --service critical-app-1
# 3. Integrate organizational knowledge
newrelic autopilot knowledge upload --from runbooks.yaml
Documentation and full API references are available through New Relic’s platform guides, outlining supported integrations and policy controls.
The big gain: real incidents triage faster, dashboards leave the "break glass" scenario, and SREs move from reactive firefighting to proactive engineering.
What are the implications of AI-driven observability for the future of site reliability engineering?
This release cements a shift: the future of SRE is an agent-augmented loop, not endless dashboard refreshes. AI-driven observability agents make it practical to automate the most draining parts of incident management—root-cause analysis, remediation prep, and context collation.
Efficiency is the first lever. Incidents are caught and routed faster, with less duplicated work and less chance of human context loss on handoff. Risks and possible remediations are surfaced ahead of human engagement.
But the change isn’t only about speed. Trust and accuracy are the new battlegrounds: teams will need to monitor how well agents align with reality, how consistently Ground Truth reflects production state, and when to override versus automate.
New Relic’s investment here sets a real floor for the rest of the ecosystem—pushing observability from a human-driven, dashboard-click-heavy exercise into a substrate for AI agents and autonomous systems. Platforms that can’t surface data cleanly, or that keep state locked in dashboards, will be left behind. Teams investing early in agent-driven observability and standardization stand to win on both resilience and maintainability.
What this enables for platform teams
The upshot is clear: New Relic Autopilot and Ground Truth move incident management away from screens and toward code. By enabling AI agents to act directly on high-quality, unified telemetry, SRE teams can scale their impact and cut through dashboard bloat.
Real outcomes: on-call pain drops, mean-time-to-resolution falls, and response consistency rises. Human engineers spend less time scavenging for fragmented incident context, and more energy on steady-state improvements.
Teams looking to future-proof their reliability practice should eval Autopilot and Ground Truth now—not just for point solution value, but as the start of the next cycle of observability and operational automation.
[Internal references: Observability best practices with OpenTelemetry, Automated incident response tools for DevOps teams, AI and machine learning in site reliability engineering, New Relic platform tutorials and API integration guides)
Top comments (0)