New Relic has launched New Relic Autopilot and New Relic Ground Truth, aimed at businesses building or using AI agents.
The two offerings represent the next stage of its platform strategy around what it calls agentic AI-first operations. One is designed to handle incident response tasks as an automated site reliability engineering agent, while the other feeds observability data into AI agents that customers already use.
Autopilot is positioned as an out-of-the-box automated SRE tool. It begins analysing an incident as soon as an alert fires, with the aim of helping teams identify root causes, assess whether action is safe and decide what to do next.
The service includes specialist agents for areas such as Kubernetes, along with tools for Kafka troubleshooting and cross-stack root-cause analysis. It also draws on an organisation's runbooks and retrospectives through New Relic Knowledge, while external Model Context Protocol connections to Jira and GitHub pull in code details.
The product also includes long-term memory to capture internal operational knowledge and make it available across teams. It can be used through a new agent-focused interface in the New Relic platform, in Slack and through automated workflow actions triggered by events such as a severe incident, a software deployment or a scheduled task.
Two approaches
Ground Truth takes a different approach. Rather than providing a managed agent, it gives customers' existing AI agents access to New Relic's observability data through tools designed for machine use rather than dashboard-based interaction.
The product works with third-party and in-house agents, including GitHub Copilot, Claude Code, AWS DevOps and custom orchestrators. New Relic says it is intended to expose richer insights than are available through public APIs or basic query tools, while reducing the number of tool calls an agent needs to find useful information.
New Relic also cited what it described as customer proof for Ground Truth, saying a large enterprise self-measured a 1.1% error rate across more than 1,300 users. It did not identify the customer.
The launch reflects a broader shift in how software vendors are adapting monitoring and operations tools for AI-driven workflows. In traditional observability products, engineers usually inspect dashboards and alerts directly. AI agents, by contrast, need data access through interfaces that can be queried programmatically and acted on automatically.
Camden Swita, Head of AI at New Relic, outlined that view in comments released with the announcement.
"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," Swita said.
"That's exactly what New Relic Ground Truth delivers: it grounds your own agents in the real truth of your systems, through APIs instead of UIs. And for teams who'd rather we run the agent, there's New Relic Autopilot, our expert agent operated for you. Both sit on the same rich data substrate, and either way the toil is reduced," he said.
Operational focus
New Relic's emphasis is on incident management and operational response, where AI tools are increasingly being applied to reduce repetitive work for engineering teams. The difficult part of an incident often lies not in detecting a failure, but in understanding its cause and deciding whether a remediation step is safe, especially in large and complex software environments.
By combining telemetry, internal documentation and links to code management systems, New Relic is seeking to place its observability platform closer to the centre of automated operations. That approach also allows it to serve two groups of customers: those that want a vendor-managed agent and those that prefer to build or orchestrate their own.
The split is significant because many large organisations are experimenting with multiple AI agents across software development, operations and support. Tools that can plug into those existing systems may be easier to adopt than products that require teams to replace their preferred platforms.
New Relic has framed both offerings around a shared data layer. In practice, that means the same underlying observability information can either be used by New Relic's automated agent or accessed by customer-run agents through dedicated interfaces.
New Relic says the result of its automated analysis is presented as a structured account grounded in system data and operational context.