At 3:17 AM, an on-call engineer wakes to a New Relic alert: the error rate on the checkout service has exceeded 5% for the last five minutes. She opens a dashboard, switches between charts, runs a NRQL query, checks logs, and finally finds a bad deployment and rolls it back. Total time: 45 minutes. During that time, thousands of customers couldn't pay.
Now imagine the same alert handled by ASI Biont. The agent detects the anomaly, fetches the deployment history through New Relic's API, correlates log patterns, flags the problematic release, and rolls it back—before the engineer finishes their first sip of coffee. That's not science fiction. It's what becomes possible when you connect New Relic to an AI agent.
In this article, I'll explain why this integration matters, what it automates, and how you can set it up in minutes—not by clicking buttons in a dashboard, but by simply telling the AI agent what to do.
Why Connect New Relic to an AI Agent?
New Relic is one of the leading observability platforms for DevOps. It ingests metrics, events, logs, and traces from your entire stack, then turns them into dashboards and alert conditions. According to New Relic's official documentation, NerdGraph is a GraphQL API that gives you programmatic access to most of this data (link). The platform answers questions like "Which service is slow?" or "When did the error rate jump?" But it often doesn't answer "Why?"—and it never automatically fixes anything.
That is the gap ASI Biont fills. Instead of having a human look at charts, the AI agent looks at the data, reasons about it, and acts on it. An observability tool shows you there is a fire; an AI agent with integration can find the source, call the fire department, and turn on the sprinklers.
What the Integration Automates
Connecting New Relic to ASI Biont turns manual toil into automated workflows. Here's a comparison of the same tasks, with and without the AI agent:
| Task | Without AI agent | With ASI Biont |
|---|---|---|
| Alert investigation | On-call engineer opens the dashboard and explores charts | Agent fetches metrics and drill-downs via NerdGraph automatically |
| Log analysis | Greps through logs, manually searches for error patterns | Agent queries the Logs API, clusters similar errors, and identifies root cause |
| Change correlation | Checks deployment timeline to match spikes | Agent combines release markers with anomaly data |
| Remediation | Runs a runbook manually, creates tickets | Agent triggers a rollback, scales a service, or sends a notification via APIs |
| Incident report | Engineer writes a summary after resolution | Agent produces a detailed post-incident report from the data |
The key point is that the integration doesn't just display data—it closes the loop. The agent can observe, reason, and, if you grant permission, take action.
Real-World Use Cases
1. High latency in a payments microservice
A payment service starts responding slowly. The alert fires in New Relic. ASI Biont:
- Queries NerdGraph for the service's ApDex score and response time percentiles.
- Pulls distributed traces and sees that a query to a Redis cluster is the bottleneck.
- Checks New Relic logs for connection pool exhaustion messages.
- Correlates this with a deployment that changed the Redis client settings an hour ago.
If you gave the agent authorisation, it can apply the fix—for example, by rolling back the configuration through your deployment API. Otherwise, it sends a detailed explanation to your team's Slack channel with an exact recommendation. The on-call engineer's job goes from "investigate for an hour" to "read a summary for two minutes."
2. Proactive anomaly detection
Many companies configure New Relic alerts for CPU and memory, but not for subtle anomalies like gradually slowing queries. ASI Biont can run scheduled checks using NerdGraph, compare current metrics against historical baselines, and detect trends before they become customer-visible incidents. If you allow it, the agent can trigger infrastructure changes through your cloud provider's API to prevent an outage. This kind of proactive AI monitoring is where modern DevOps observability is heading.
3. Multi-service incident correlation
A modern application often has dozens of microservices. A failure in one can cause cascading errors in others. ASI Biont can query New Relic's entity relationship data, correlate alerts from multiple services, and determine the single source of failure. Instead of receiving five separate pages, your team gets one message: "The cart service is the root cause. The order and payment services are failing because their dependency is unhealthy." That reduces alert fatigue significantly.
How the Connection Works: A Chat, Not a Dashboard
Here's the part that surprises most engineers. ASI Biont does not have a "New Relic integration" button in some admin panel. It connects through conversation.
You simply open a chat with the agent and say:
"Connect to New Relic. My API key is NRAK-XXXX."
The agent then writes the integration code on the fly. It reads New Relic's official API documentation, creates the necessary authentication wrapper, and confirms: "Connected. I can read account metrics, alerts, logs, and deployment markers. Which incidents do you want me to handle automatically?"
That's it. No dashboard buttons, no "Add Integration" wizard, no waiting for the development team to release an adapter. The user just provides the API key, and the AI builds the integration instantly.
This approach is not limited to New Relic. ASI Biont can connect to any service with an API. Whether it's your cloud provider, a ticket system like Jira, or a messaging platform like Slack, the agent writes the integration in real time. This means your tooling doesn't have to be "officially supported"—if it has an API, it can be connected. You no longer need to wait for vendors to ship native integrations; the AI agent becomes the universal adapter.
Security by Default
When you paste an API key into chat, it's not just a string—it's a credential. ASI Biont treats API keys as secrets and never exposes them in plain text after connection. It uses the key only for the specific integration you requested, and you can revoke it in New Relic at any time. For extra safety, we recommend creating a scoped API key that only has read access if you want the agent to analyse but not change anything. For automation, you can grant restricted write permissions to specific services. This way, you get the benefit of AI-driven incident response without giving up control.
Built on Official APIs
This integration is built on New Relic's official public endpoints:
- NerdGraph API for querying metrics, alerts, and deployments: https://docs.newrelic.com/docs/apis/nerdgraph/get-started/introduction-new-relic-nerdgraph/
- Logs API for querying logs and log patterns: https://docs.newrelic.com/docs/logs/log-management/log-api/introduction-log-api/
- Alerts API for managing notifications and policy details: https://docs.newrelic.com/docs/alerts-applied-intelligence/new-relic-alerts/get-started/introduction-to-alerts/
Using official APIs means the agent gets the same data that powers New Relic's own UI, but in a structured format it can reason over. The code generated by ASI Biont is unique to your account and your workflows; it is not a generic plugin. Each integration is written on the spot, so it fits your exact use case.
Why It Pays Off
The biggest savings are time and attention. Most SRE teams report that alert fatigue and manual investigation are their top productivity killers. With an AI agent taking over triage, your people only get involved when action is genuinely needed—and when they do, they already have a clear summary of the root cause and suggested next steps.
Imagine never again waking up to a vague alert. Instead, the message reads: "Incident resolved. The checkout service was rolled back to version 2.14 due to a memory leak in the payment client. Error rate declined to 0.5% within 9 minutes. Full incident report attached." That's the level of automation this integration is built for.
The routine part of incident response—gathering data, running queries, checking logs, correlating changes—is exactly what an AI agent does best. Humans are better at making nuanced decisions and handling the unexpected. By automating the routine, you free your engineers to focus on the interesting problems that actually move the business forward.
From Alert to Resolution—Try It Now
Connecting New Relic to ASI Biont is a five-minute experiment that will change how you think about DevOps observability. There is no complicated setup, no waiting for an integration catalog, and no need to learn a new interface. Just open a chat at asibiont.com, paste your New Relic API key, and ask the agent to watch your alerts. You'll see it analyse real data, explain what's happening, and suggest actions. You can keep it in read-only mode first, and when you trust the results, give it the ability to act.
Your future on-call self will thank you.
Comments