Insights are only half the job

Open 360 AI chat is good at answering the hard questions: why an alert fired, what changed, which errors matter. But every investigation ends with the same kind of follow-up work:

• “We should have an alert on this.” The error pattern the agent found deserves monitoring, so it doesn’t surprise you twice.
• “We should have a dashboard for this.” The metrics and queries the agent assembled during the investigation are exactly the view the team needs next time.
• “Someone should clean this up.” Noisy alerts need tuning, stale ones need retiring.

Until now, that work left the conversation. The agent handed you its findings, and you opened another tab to build the alert or dashboard by hand, translating what the agent already knew into clicks and configuration. The knowledge was there; the follow-through was manual.

From advisor to operator

Agentic observability means agents that don’t stop at describing your system. They should help you improve it. Creating observability assets is the natural next step, and it changes what a conversation with Open 360 AI can accomplish:

• Investigations end with coverage, not just conclusions. The agent that found the failure pattern can create the alert that catches it next time, with the query, threshold, and context it already worked out.
• Dashboards are built from understanding. A dashboard created at the end of an investigation reflects what actually mattered: the failing service, the signals that moved, the numbers worth watching.
• The loop closes where it opened. You asked the question in chat. The answer, and the durable asset that comes out of it, land in the same conversation.

This is the same principle behind our integrations expansion: context makes agents smarter, and the ability to act makes them useful. Here, the action lands in the platform itself.

What the agent can create

The new skills teach Open 360 AI chat the Logz.io public APIs, across the Open 360 platform. Alerts and dashboards are where investigations most naturally lead:

• Alerts. Create and update log and metric alerts: the query, the conditions, the severity, the notification targets.
• Dashboards. Create and update dashboards, so the view the agent assembled during an investigation becomes a permanent one.
• Supporting objects. The agent can also manage the pieces around them, such as notification endpoints and saved searches, so the assets it creates are complete and wired up.

And because the agent can read the same APIs, it can work with what you already have: list your alerts, spot gaps or duplicates, and propose changes before making them.

How it works

Open 360 AI chat uses the same guarded integration mechanism that powers its 600+ external integrations, now pointed at Logz.io’s own public APIs. The agent knows what each API offers and how to use it well. When you ask for an alert or a dashboard, it composes the object from the context of your conversation and creates it through the same APIs your team would use.

Access is deliberately conservative:

• Read-only by default. Out of the box, the agent can look but not touch.
• Write is an explicit admin decision. An account admin enables write access in Open 360 AI settings. Until then, the agent’s write capabilities are not exposed at all.
• Scoped credentials. The agent operates with deliberately limited tokens, following the same least-privilege approach we recommend for every integration.

You are always the one asking. The agent creates what you told it to create, in your account, through the platform’s public APIs, and the result is a normal Open 360 object you can inspect, edit, or delete like any other.

A concrete example: from investigation to coverage in one conversation

The investigation. You ask Open 360 AI chat why checkout latency spiked overnight. The agent works through the logs and metrics and isolates the pattern: a burst of timeout errors from one downstream dependency, starting at 02:10.

The gap. You ask, “would we have caught this earlier?” The agent checks your existing alerts and finds none covering this error pattern; the closest alert watches overall error rate, which never crossed its threshold.

The fix, in the same chat. You say: “Create an alert for this.” The agent already has everything it needs from the investigation: the query that isolates the pattern, a sensible threshold based on the night’s data, and a severity to match the impact. It creates the alert and shows you exactly what it configured.

The view. One more message: “And give me a dashboard for checkout health.” The agent builds it from what the investigation surfaced: the latency percentiles, the error pattern it isolated, the dependency that caused the trouble.

Few minutes earlier, none of this existed. There was an unexplained spike and an open question. Now there’s an explanation, an alert that catches the next occurrence, and a dashboard the team can stand behind, all from one conversation, without leaving chat.

The road ahead

Creating alerts and dashboards is the first step of Open 360 AI operating the platform on your behalf. The direction is an agent that not only investigates and explains but continuously helps keep your observability posture healthy: proposing coverage where it’s missing, tuning what’s noisy, and retiring what’s stale, always with you in command.

For further information, feel free to reach out to our Logz.io support team at help@logz.io.

Want to see an investigation turn into an alert in one conversation? Book a demo or start a free trial today.

‍

FAQS

Can the agent change things without me asking?

No. The agent creates or changes objects in response to your requests in chat. And unless an account admin has enabled write access, it can’t make changes at all.

How do I enable write access?

An account admin turns it on in Open 360 AI settings. It’s off by default, and you can turn it back off at any time.

What can the agent create today?

Open 360 objects available through the Logz.io public APIs, including log and metric alerts and dashboards, plus supporting objects like notification endpoints and saved searches.

What if the agent creates something wrong?

Everything it creates is a standard Open 360 object in your account. You can review, edit, or delete it exactly as if a teammate had created it, and the agent shows you what it configured when it’s done.

Does this use the same permissions as my integrations?

It follows the same least-privilege approach: scoped credentials, read-only by default, and write only behind explicit admin consent.

Does this work for metrics and SIEM too?

The skills cover Logz.io’s products through their public APIs, with each product’s capabilities available on accounts provisioned for it.

‍

‍