Skip to content

1.1. What the CogentAI widget is

What it does for you — ask, and it queries, analyses and interprets

The CogentAI widget is a chat window that floats on top of the console. Ask it something and it answers.

The difference from an ordinary chatbot is that it does not recite what it learned in advance. It queries your systems directly and answers from what it finds.

What happens when you ask​

Suppose you ask "How was <app name>'s response time yesterday afternoon?" (Replace <app name> with a name that actually appears in your console.)

What the widget does when you ask — interpret the question, query metrics, calculate, respond

To do step ② by hand you would open a dashboard, set the time range and pick the metrics. The widget does that for you.

What you can ask about​

SubjectExamples
Metrics"How is <app name>'s response time?" · "When will the node disk fill up?"
Kubernetes"Why won't <pod name> start?" · "What does this event mean?"
Databases"Why is this query slow?" · "Show me the table structure"
Logs"What does this log line mean?"
How to use the product"How do I create an alert rule?"

The last one is not a query. It searches the documentation and answers from it, so you can ask instead of digging through the product manual.

What comes with the answer​

You get more than text.

  • Charts — metric trends, forecast bands, execution plan trees
  • Tables — lists and statistics
  • Diagrams — relationships between tables (ERD), structure diagrams

Charts can be zoomed or opened full screen (Chapter 203).

Three things to know before you start​

One. There are buttons all over the console​

Look for a button before you start typing. Charts, resources, logs and SQL statements each have an AI button next to them, and pressing it hands over exactly what you were looking at.

That is faster and more accurate than retyping. Part 6 covers where the buttons are in each console.

Two. It can be wrong​

This notice is always shown at the bottom of the widget.

"CogentAI can make mistakes. Please double-check responses."

Check numbers and any command it tells you to run in particular. The charts and tables it based the answer on come with it, so read the figures too, not just the prose.

Three. It can change things, not only read them​

Most of what it does is read and answer. Within the scope your administrator has opened, however, it also makes changes: creating Kubernetes resources, triggering Jenkins builds, writing to a database that is not read-only.

That is why it asks once more when your wording involves deleting, stopping or restarting.

Dangerous request detected — do you want to continue?

Words such as delete, remove, terminate, stop and restart bring up this dialog. Press Continue if that is what you meant, or Cancel if it is not (Chapter 306).

What it is allowed to do differs from one installation to another. It depends on which tools your administrator connected. Some sites allow queries only; others allow changes as well.

The same widget in three consoles​

OPENMARU APM, COP Console and Observability all use the same widget. What differs is where the buttons are and what is handed over with your question.

Learn it in one console and you can use it in the others as they are.

Next​