The Best IT Incident Is One That Never Happens. How New Relic Keeps ServiceNow’s CMDB Ready

One number from ServiceNow's Autonomous Workforce launch stopped people mid-scroll: 90% of internal IT requests resolved autonomously. The implication was immediate. What if IT operations could largely run itself? Can any organization actually get there, or only one with 20 years of operational intelligence, purpose-built infrastructure, and domain expertise baked in?

The answer comes down to one thing: the data model the AI is reasoning from. Modern infrastructure moves faster than any manual process can document. Dependencies shift with every deployment. Topology reshapes with every autoscaling event. The relationship between two critical services exists in one engineer’s memory, and nowhere else.

For SRE teams, that gap shows up at 1am when it takes ninety minutes to reconstruct a dependency topology from memory to find a fix that takes three minutes to apply. For IT operations leaders, it shows up in a boardroom with an autonomous operations vision that depends on AI specialists acting on a service model that isn’t current enough to trust. Same gap. Two different ways it costs the organization, and one place it needs to be solved. That’s the problem the New Relic Service Graph Connector for ServiceNow is built to address.

The Gap To Fill For Autonomous Operations

AI specialists operate on data. Specifically, a real-time, accurate model of what’s running in the environment, what depends on what, and what normal looks like. Without that, autonomous decisions aren’t intelligent. They make fast guesses at scale.

Most enterprises are walking into autonomous IT operations with a CMDB that’s already out of date. 

In a cloud-native, microservices-driven environment, where topology is shifting with every deployment and every autoscaling event, manual discovery and periodic sync can’t keep pace. By the time an AI specialist needs to act, the map reasoning may already be wrong. A relationship that has existed for two years simply isn’t there. And the AI has no memory to fall back on.

That’s what stops autonomous operations before it starts. Not the AI. Not the workflows. The data foundation those things depend on.

Real-time intelligent observability closes that gap. When live entity graphs continuously enrich the CMDB (mapping services, their dependencies, their health state, and their change history), the service model becomes a living artifact rather than a historical record. Every deployment, every autoscaling event, every new dependency captured automatically and continuously reconciled against what ServiceNow knows. The map stays current because it’s fed by what’s actually running. That’s what makes autonomous decision-making trustworthy enough to act on.

The New Relic Service Graph Connector for ServiceNow is built for this. It’s two functions working together:

Continuously bringing New Relic’s live entity data into ServiceNow’s CMDB as Configuration Item records, so the service model reflects what’s actually running in production.

Routing alerts from New Relic into ServiceNow with those alerts already associated with the correct CI records. Most alert integrations fire a notification and stop there. The Service Graph Connector ties the alert directly to the enriched CI, so when something fires, ServiceNow already knows what it is, what it connects to, and what’s at risk.

Since the last release, the connector has also deepened the dependency picture by adding service-to-service relationship mapping, host-to-application relationships, and expanded application data coverage. The graph the AI specialist reasons from gets more complete with every update.

The connector is also part of a broader agentic integration story. New Relic and ServiceNow have been developing agent-to-agent capabilities that enable AI to automatically assess incident severity, identify affected services, and route to the right owners, without waiting for human triage. The accurate, continuously updated service graph the connector provides is what makes that kind of agentic reasoning trustworthy enough to act on.

Teams in financial services and healthcare are already running the connector in production. That’s early evidence that ITSM and observability are converging in the places where operational risk is highest.

Where This Is Heading

The AI specialist gets smarter as the map gets better. Consider what becomes possible when the dependency relationship between authentication and payment processing is known. Not from a Confluence page, not from an engineer’s memory, but from a live service graph continuously updated by real observability data. Blast radius assessment becomes automatic. Because scope is understood, the AI specialist acts rather than escalates. Because every decision is logged against a verified data model, the audit trail is complete.

Ninety minutes of human investigation. Forty seconds of autonomous reasoning. Zero pages.

ServiceNow’s Autonomous Workforce is the operating model. Observability-driven CMDB accuracy is the prerequisite. Together, they describe an IT organization where the best incident is one that never happens, and where the ones that do are resolved before most people know they exist.

The teams that lead this transition will be the ones who invested in the data foundation that makes AI trustworthy enough to act without asking permission. The autonomous era of IT operations starts with better data.

The next step is making that data stream in true real time, so that agentic workflows can act on ground truth the moment something changes, not the moment someone last ran a sync. That’s the direction New Relic and ServiceNow are building toward.