AI assistant

Ask, configure and act in plain language, from the desktop console, the web console or the command line. Runs on OpenAI, Anthropic Claude or a self-hosted Ollama model, so data can stay on your network.

01

Dashboards from a prompt

"Build a dashboard for web-01 with CPU and memory line charts and an alarm viewer"

The assistant creates the dashboard element by element and validates each one as it goes.

Dashboard generated by the assistant: CPU and memory line charts and an alarm viewer for web-01
Incident analysis answer from the assistant, with topology context and correlated alarms
02

Ask about anything in the system

  • "Why is the core switch down?"
  • "Explain this alarm."
  • "Show me a chart of interface errors on this node."
  • "What changed in the logs last night?"

Incident analysis gathers topology context, correlated alarms and history. Log analysis searches syslog, Windows events and traps together.

03

Configure by describing

Create and edit metrics, set thresholds, add event processing rules, open and close maintenance windows.

Threshold created by the assistant from a plain-language request
Approval prompt shown before the assistant runs an action
04

Act, with approval

In the open-source core: maintenance mode, on-demand device configuration backup, notifications, web service calls.

Remote operations on hosts and devices, such as restarting services, running commands over SSH or deploying packages, are Enterprise skills. Editions

Where it runs

  • Desktop console
  • Web console
  • nxai command line

Models

  • OpenAI
  • Anthropic Claude
  • Self-hosted Ollama