New Relic Bets on the Plumber Role in the Agentic AI Stack

Picture a busy restaurant kitchen where every new hire wants to be head chef. New Relic has just walked in and volunteered to run the dishwashers. On 24 February 2026 the vendor laid out a preview roadmap that deliberately refuses to chase the full software delivery lifecycle, choosing instead to be the clean-dishes layer that every other agent depends on. In a market where Datadog, Dynatrace and Cisco-Splunk are all angling for the chef's hat, that is either humility or strategic judo.

What Happened

New Relic announced a batch of agentic AI features, all in preview until next quarter, that push AI agents deeper into its observability and AIOps platform. As TechTarget reported, chief product officer Brian Emerson framed the strategy as explicit: expand agentic support, but work with partners in adjacent markets rather than moving onto their turf.

The headline act is a new SRE Agent that runs incident investigations across observability data and cooperates with incident management tools from ServiceNow, Atlassian and PagerDuty. Notably, New Relic is not building an AI agent control plane for incident remediation that would overlap with those partner tools. The dishwasher stays in its lane.

For the software delivery side, Emerson pointed at CI/CD partners including Amazon and Microsoft's GitHub. "When we've identified a problem with a specific change that happened in the infrastructure or configuration, we can push that information back into an agent on the CI/CD side," he said. "It used to take hardcoded API integrations to get that work done, and now it's just dynamic and fluid."

There is also a new no-code AI agent builder, which Emerson is careful to say is not a replacement for general-purpose agent builders. Around all that, New Relic is adding AI agent monitoring with business impact analysis for AI applications, federated logs on Amazon S3, eBPF network metrics, user notebooks, and dynamic homepages replacing the old dashboard model. Most of those capabilities, the piece notes, already exist in some form at Cisco-Splunk, Dynatrace and Datadog.

Technical Anatomy

The guts of it: New Relic is betting that agent-to-agent protocols will replace hand-rolled API integrations as the connective tissue between observability, ITSM and CI/CD. That is the bit Emerson is really selling when he says the agents are "smart enough to figure out how to interact." In practical terms, the SRE Agent becomes a knowledge endpoint that a ServiceNow agent or a GitHub-side agent can query for root cause context, user impact, or the specific config change that broke production at 4pm on a Friday.

Anyone who has wired up a Jira-to-PagerDuty-to-Splunk integration with bespoke webhooks knows the pain being targeted here. Every version bump on either side breaks the glue. If agentic mediation works, the contract becomes a semantic one: "tell me what changed in this service" instead of "POST to /api/v3/incidents with this exact schema." That is a meaningful shift for platform teams who currently spend unglamorous weeks maintaining connector code.

The business impact analysis for AI applications is the other technically interesting bit. Monitoring LLM-backed services is not the same as monitoring a stateless REST API. Token costs, prompt drift, hallucination rates and retrieval quality all need to land somewhere in the observability stack, and the standards are still being argued over in OpenTelemetry working groups. New Relic slotting AI workload telemetry next to traditional metrics and logs is table stakes at this point, not differentiation.

The federated logs on S3 feature deserves a note. Rather than forcing customers to ship every log byte into New Relic's backend, federation means querying where the data already lives. That is the direction the whole category is moving, because storage egress and ingest pricing have become the part where it all falls over for large customers. Add eBPF network metrics, which collect kernel-level telemetry without sidecars or code changes, and you have a platform catching up to where the Datadogs of the world already parked two years ago.

Who Gets Burned

The awkward truth Emerson himself admits: "There are no clear winners in the race around AI. It's still new and being kind of defined in a lot of different ways." Stephen Elliot, an analyst at IDC, points out that New Relic, Datadog and Dynatrace all have large customer bases and all have had to retrofit their tools for generative and agentic AI. Customers, he says, are interested but skeptical. "That looks intriguing, that sounds really great, but I need to see it."

New Relic's exposure is strategic. Torsten Volk, an analyst at Omdia, puts it bluntly: "New Relic is doubling down on the operator persona while most of the competition is racing to own the full software delivery lifecycle." In an open agentic world, that focus might age well. In a consolidation-driven procurement cycle, "however, being an 'additional layer' in the stack is risky, as customers might prefer to just get one platform." That is the squeeze.

For iGaming and fintech platform teams, this matters directly. If your incident stack already runs on PagerDuty or ServiceNow and your pipelines on GitHub, New Relic's "we integrate, we don't replace" pitch is the path of least resistance. If you are a crypto or DeFi shop that has consolidated onto Datadog because the CFO got tired of six invoices, the pitch lands differently.

The other burn risk is internal. New Relic was acquired by a private equity firm in 2023 and went through an executive reshuffle the same year. PE ownership tends to shorten patience for strategies that take three years to prove out. The operator-persona bet is exactly that kind of strategy. If the next four quarterly reviews show Datadog winning full-SDLC deals, the board conversation gets interesting.

Playbook for Engineering Teams

If you are a platform lead or SRE manager staring at this news, three moves make sense this week.

First, audit your existing observability-to-ITSM glue code. Count the bespoke API integrations between your monitoring tool, your incident management platform and your CI/CD system. That number is your migration cost if agentic mediation actually works as advertised. It is also your use in any renewal conversation with your current vendor.

Second, get serious about AI workload telemetry before procurement forces your hand. Whether you end up on New Relic, Datadog or Dynatrace, you need a clear spec for what an LLM-backed service must emit: token counts, latency percentiles, retrieval hit rates, failure modes. Write it down now, pin it to OpenTelemetry semantic conventions where they exist, and you will not be at the mercy of whichever vendor ships the shiniest demo.

Third, pressure-test the "agents collaborate across vendors" promise. Ask your New Relic rep to demonstrate, on your data, an end-to-end flow where the SRE Agent answers a question posed by your ServiceNow instance or your GitHub pipeline. Preview features are exactly that. Elliot's customers are right to say "I need to see it." You should too, because buying the roadmap instead of the product is how platform teams end up with a two-year regret.

For teams already happy with their current stack, the honest move is patience. Let the preview become GA, let the competitors respond, and revisit in two quarters when pricing and real-world reliability data exist.

Key Takeaways

• New Relic is deliberately staying out of the full software delivery lifecycle and betting on partner integrations with Amazon, GitHub, ServiceNow, Atlassian and PagerDuty instead.
• The new SRE Agent, no-code agent builder and AI application monitoring are all in preview until next quarter, so production bets are premature.
• Federated logs on S3 and eBPF network metrics bring New Relic closer to parity with Cisco-Splunk, Dynatrace and Datadog, not ahead of them.
• Analyst consensus is that no vendor has won the agentic observability race yet and enterprise buyers are demanding proof, not pitches.
• The operator-persona strategy is defensible in an open agentic future and dangerous in a consolidation cycle. Watch the next two quarters of deal flow closely.

Back to the kitchen. Running the dishwashers is not glamorous, but every chef in the building needs clean plates. New Relic is wagering that when the agentic dust settles, the platform that reliably answers "what just broke and who cares" will be quietly indispensable, even if it never gets the star on the menu. The bet is rational. Whether PE patience and enterprise procurement agree is the part where it all falls over, or doesn't.