New Relic is making a notable shift with its Agentic Platform (preview announced February 2026): rather than relying on “classic IT automation” with scripts and RPA, the vendor bets on AI agents that can reason and chain operations to assist your Ops/SRE teams.
Concretely: prebuilt agents (including an SRE Agent), an iRCA (intelligent root-cause analysis), and native integrations with ServiceNow, Atlassian, PagerDuty via the Model Context Protocol. The promise: less time triaging, more time fixing and improving. This is the kind of evolution that speaks to CTOs and founders of SMEs and mid‑market enterprises that run small, already stretched IT teams.
The SME opportunity
The headline number is hard to ignore: engineers reportedly spend 33% of their time reacting to incidents. For a run team of 3–6 people, that quickly adds up to many staff-weeks per year wasted chasing symptoms across logs, metrics, traces, tickets and Slack.
The value of an agentic approach is that it’s not a brittle automation. An RPA or script does exactly what it was told—until the environment changes (new dependency, changed log format, altered escalation path) and it breaks. An AI agent can reason across multiple steps: correlate signals, propose a root-cause hypothesis, open or update tickets, suggest remediation, and learn from your practices.
Another SME-friendly angle: no-code interfaces and prebuilt agents. That reduces the need to pull engineers off feature work just to wire tools together. In short: faster adoption and an easier ROI to justify.
Caveats & watch‑outs
First and foremost: vendor lock‑in. If your incident workflows, rules, automations and agents live inside New Relic’s ecosystem, reversing course can be costly—in time, rework and required skills.
Second: moving from “simple” to “complex” is not automatic. Yes, starting with a prebuilt SRE Agent is fast. But once you require robust scenarios (chained actions, exception handling, human validation, escalation flows, environment separation), you need rigorous testing and validation. Otherwise you risk industrializing poor decisions.
Third: governance. New Relic notes built‑in RBAC and audit logging. Good. But you must decide the agent’s scope: suggest only? execute commands? open tickets? trigger automated remediations? Poorly scoped autonomy turns a remediation into a domino effect.
Finally: cost. Pricing is not published, and the agentic layer is presented as a separate cloud component—meaning potential additional charges on top of existing observability. Validate total cost of ownership before committing.
Compliance and data considerations
This is where data matters: logs, traces, metrics, application topology—and sometimes user identifiers, business errors, or items that qualify as personal data. An iRCA that correlates across a graph can therefore process sensitive data depending on your context.
- Residency: New Relic is headquartered in San Francisco. Verify where your telemetry is collected and stored (is an EU region available?)—without an appropriate EU region the GDPR risk increases.
- DPA: insist on a Data Processing Agreement compliant with GDPR (and Switzerland’s nLPD if you serve Swiss customers).
- EU AI Act: if an agent takes autonomous decisions that impact business processes or compliance, qualify the use case and tightly define its autonomy.
- Alternatives: depending on constraints, evaluate EU‑oriented options (for example Scaleway Observability, and check offerings from Infomaniak or OVH).
Conclusion & Cohesium support
AI agents applied to observability can be a significant accelerator: less triage, better availability, and an IT organization that redeploys time toward long‑term improvements. But between lock‑in, governance, compliance and opaque pricing, the question isn’t “is this cool?”—it’s “is this controllable and cost‑effective for our organization?”
Rather than cobbling solutions together, Cohesium AI can support you with: (1) an AI strategy audit (agents vs RPA, governance, roadmap), (2) a Compliance & Data audit (GDPR/nLPD, data residency, DPA, EU AI Act impact), and (3) bespoke development when you need to go beyond vendor templates and avoid vendor lock‑in.