AI Assistant: Getting Started

This tutorial walks you through setting up the AI assistant from scratch and verifying it works.

For how-to guides on specific tasks, see AI Assistant How-to Guides. For configuration reference, see AI Assistant Reference.

Prerequisites

  • A running NetXMS server (version 7.0 or later).

  • Access to at least one LLM provider: a cloud API (OpenAI, Anthropic, Mistral) or a local Ollama instance.

  • An API key or token for the chosen provider (not required for Ollama with default settings).

  • At least one server module that provides AI skills and functions. The aitools module is included in the open-source edition. Load it by adding Module=aitools to netxmsd.conf (before any named section).

Step 1: Configure an LLM Provider

Add a provider section to the server configuration file (netxmsd.conf). The example below configures an OpenAI-compatible provider:

Module=aitools

[AI/Provider/openai]
Type = openai
URL = https://api.openai.com/v1/chat/completions
Model = gpt-4o
Token = sk-your-api-key-here
Slots = default

Restart the NetXMS server after changing the configuration file.

The AI assistant is only available when at least one provider is configured. The default provider is designated either by assigning the default slot to a provider or by setting DefaultProvider in the [AI] section. The AI subsystem is also disabled when no AI functions or skills are registered (i.e., no module providing them is loaded); the server itself starts normally in all cases.

Step 2: Open the AI Chat

In the management client:

  1. Switch to the AI perspective. It appears in the Administration section of the perspective switcher when the AI component is active on the server.

  2. Open the AI Assistant view (the perspective also provides the AI Messages and Observations views).

  3. Type a message in the input field at the bottom and press Enter.

The AI assistant will respond using the configured LLM and may call internal functions to retrieve live data from the server (e.g., alarm lists, object details, historical metrics).

Step 3: Ask a Question

Try a simple query to verify the setup:

What are the current active alarms?

The AI assistant will use the alarm-list function to retrieve real-time alarm data from the server and present the results. Function calls are displayed in the chat as progress indicators so you can see what data is being accessed.

Next Steps