New Relic Agentic Platform
by New Relic
No-code builder, runtime and governance for observability AI agents, including the Autopilot SRE agent
New Relic's agentic layer for its observability platform: a no-code builder and runtime for custom operations agents, plus Autopilot, a New Relic-operated SRE agent that triages alerts, finds root causes and proposes fixes. It is aimed at SRE and platform teams already sending telemetry to New Relic who want less manual incident toil.
New Relic Agentic Platform is the umbrella New Relic uses for building, running and governing AI agents on top of its Intelligent Observability Platform. It was announced in preview on February 24, 2026 with four parts: a drag-and-drop no-code agent builder meant for SREs and operations leaders rather than ML engineers, prebuilt agents such as SRE Nerd, a dynamic agent runtime for multi-step reasoning, and a unified orchestration console, with Model Context Protocol (MCP) support for tool access, role-based access control, audit logs, a built-in evaluation engine and native hooks into the generally available Workflow Automation product. The flagship agent has changed name and status since launch. In March 2026 the SRE Agent gained Intelligent RCA, which searches the entity topology graph with probabilistic causal models, and on June 23, 2026 New Relic introduced Autopilot, which triages an incident when an alert fires, identifies root causes and scopes remediations, with specialist agents for Kubernetes and Kafka, grounding in an organisation's runbooks and retrospectives, long-term memory, Slack access and MCP connections to Jira and GitHub. Alongside it, Ground Truth exposes agent-optimised observability data to outside agents such as GitHub Copilot, Claude Code and AWS DevOps. New Relic's documentation records the SRE Agent being renamed Autopilot and reaching general availability on July 24, 2026, with release 2.1 in August adding a GitHub MCP integration in public preview and release 2.2 in September adding parallel investigation steps and GitHub Enterprise Cloud support. Autopilot requires the Advanced Compute add-on, is limited to one agent and 100 requests per hour per organisation, and executes runbook actions only with human confirmation. New Relic has been privately owned by Francisco Partners and TPG since November 2023.
SRE and platform engineering leaders at organisations already standardised on New Relic, because the agents act on telemetry New Relic already holds and are billed through its existing Advanced Compute add-on.
Faster incident triage: an agent starts investigating the moment an alert fires, correlating logs, metrics, traces and changes into a proposed root cause and next steps.
At a Glance
- Category
- AI Agents & Orchestration
- Pricing
- Usage-based, Subscription, Contact for pricing
- Target Market
- CTOs, VP Engineering, SRE Leaders, Platform Engineering Teams
- Deployment
- Cloud-only, API-based
- Founded
- 2008
- Headquarters
- San Francisco, United States
- Team Size
- 500+
Key Features
- ✓No-code agent builder
Drag-and-drop designer that lets SREs and operations leaders turn institutional knowledge into agents without writing code or involving ML engineers.
- ✓Autopilot SRE agent
Triages incidents when an alert fires, identifies root causes and scopes remediations, with Kubernetes and Kafka specialist agents built in.
- ✓Intelligent RCA
Searches the entity topology graph with probabilistic causal models, combining deterministic analysis with generative AI to narrow likely root causes.
- ✓Ground Truth for third-party agents
Gives GitHub Copilot, Claude Code, AWS DevOps or custom orchestrators token-efficient, agent-optimised access to New Relic observability data.
- ✓Governance and evaluation
Role-based access control, audit logs and a built-in evaluation engine that continuously tests agent performance so operators can control what agents do.
- ✓Knowledge and memory
Grounds conclusions in an organisation's own runbooks and retrospectives and keeps long-term memory of tribal operational knowledge.
- ✓Human-confirmed remediation
Runbook actions run only after explicit human confirmation, keeping production changes behind governance guardrails.
Capabilities
Use Cases
- •Alert triage for on-call engineers
When an alert fires, Autopilot investigates the telemetry and shares a root-cause hypothesis and suggested next steps with responders in Slack.
- •Kubernetes and Kafka troubleshooting
Specialist agents investigate Kubernetes and Kafka incidents across the stack, narrowing the search space before engineers dig into individual services.
- •Code-aware investigations
MCP connections to GitHub and Jira pull code details and tickets into the investigation, helping pin a problem to the change behind it.
- •Observability data inside coding agents
Engineers working in Claude Code or GitHub Copilot query New Relic data through Ground Truth instead of switching to dashboards while debugging.
- •Automating operational runbooks
Domain experts build no-code agents and Workflow Automation flows that execute approved runbook steps with human confirmation, reducing repetitive toil.
Ideal For
Best For
- ✓Automated first-pass triage of alerts for on-call SRE teams
- ✓Root-cause analysis across Kubernetes and Kafka estates already instrumented in New Relic
- ✓Encoding runbooks and past retrospectives into agents that ground their recommendations
- ✓Feeding New Relic telemetry to existing coding agents such as GitHub Copilot or Claude Code via Ground Truth and MCP
- ✓No-code automation of repetitive operational workflows by domain experts
Not Ideal For
- ✗Teams not already on New Relic: the agents reason over New Relic telemetry, so the value depends on first moving observability data there and paying for its ingest.
- ✗Organisations on New Relic's free or original product-based pricing: Autopilot requires the Advanced Compute add-on, whose per-unit price is not published.
- ✗Buyers wanting a general-purpose enterprise agent platform: New Relic's chief product officer told TechCrunch it is built for observability outcomes rather than general agent management.
- ✗Teams that need predictable spend: Capterra and PeerSpot reviewers of the wider New Relic platform repeatedly cite high, usage-driven costs.
Deployment
Market Analysis
Pros
- ✓The out-of-the-box SRE agent (Autopilot) is generally available, not just a preview
- ✓Recommendations are grounded in runbooks, retrospectives and topology data rather than free-form model output
- ✓Human confirmation, RBAC and audit logs on agent actions suit regulated production environments
- ✓Open to third-party agents through MCP and Ground Truth, so it complements existing coding assistants
Cons
- ✗Only as useful as your New Relic footprint: agents reason over New Relic telemetry, so non-customers face a platform migration first
- ✗Opaque cost: Advanced Compute CCU prices are unpublished, and CCUs are metered by how often features are used rather than by data volume
- ✗Hard limits in the current release: one Autopilot agent and 100 requests per hour per organisation, which New Relic's own docs say calls for a phased rollout
- ✗Reviews of the wider New Relic platform on Capterra (4.5/5, 199 reviews) and PeerSpot (4.2/5, 176 reviews) repeatedly cite high, usage-driven pricing and a steep learning curve
- ✗Much of the stack is young: the no-code builder launched in preview in February 2026 and the GitHub MCP integration was still in public preview in August 2026
Pricing
Standard
From $10/mo (first full platform user)
- ✓$99 per additional full platform user, maximum of 5
- ✓Core users $49/user/month
- ✓100 GB ingest free, then $0.40/GB
Pro
$349/full platform user/mo (annual)
- ✓$418.80/user on monthly pay-as-you-go
- ✓Unlimited full platform users
- ✓2-hour critical support SLA
Enterprise
Contact for pricing
- ✓FedRAMP Moderate and HIPAA eligibility with Data Plus
- ✓1-hour critical support SLA
Advanced Compute add-on (required for Autopilot)
Contact for pricing
- ✓Metered in compute capacity units per successful action
- ✓Compute budgets with threshold alerts
- ✓Features toggled on or off in Feature Control Manager
New Relic does not price the agentic features on their own. Autopilot requires the Advanced Compute add-on, which is metered in compute capacity units (CCUs) charged per successful customer-initiated action; CCU prices are not published and sales quotes them by edition. That comes on top of the normal bill: per-GB ingest ($0.40/GB after 100 GB free, $0.60/GB on Data Plus) and full-platform user seats ($99 per additional user on Standard, $349 on Pro with an annual commitment). Advanced Compute is not available on free or original product-based pricing.
Security & Compliance
Connect
Sources
This page was written from 16 sources, 11 on domains other than newrelic.com.
- 1.newrelic.com — 20260224 1vendor
- 2.newrelic.com — 20260623 2vendor
- 3.newrelic.com — sre agentvendor
- 4.newrelic.com — pricingvendor
- 5.newrelic.com — computevendor
- 6.docs.newrelic.com — sre agent setup
- 7.docs.newrelic.com — docs release notes
- 8.docs.newrelic.com — regulatory audits new relic services
- 9.techcrunch.com — new relic launches new ai agent platform and opentelemetry t
- 10.itbrief.com.au — new relic unveils agentic ai platform for sre automation
- 11.itbrief.com.au — new relic launches autopilot ground truth for ai
- 12.helpnetsecurity.com — new relic agentic platform
- 13.apmdigest.com — new relic updates sre agent
- 14.peerspot.com — new relic pros and cons
- 15.capterra.com — reviews
- 16.en.wikipedia.org — New Relic
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Relevance AI
No-code AI agent builder for sales, marketing, support and ops teams to build and run a multi-agent AI workforce
Capacity
AI-native customer experience platform: omnichannel AI agents, real-time agent assist and automated QA on one knowledge layer
xpander.ai
Vendor-neutral enterprise AI agent platform with a universal agent harness and Omni, an AI forward-deployed engineer
WSO2 Agent Manager
Open-source control plane for governing AI agents across any framework, model or cloud