Features of the IBM Network Intelligence

Edit online

Network Intelligence is an AI native system that transforms complex network data to actionable intelligence that accelerates decisions, reducing downtime, and redefining network management at scale. Network Intelligence transforms your network operations with intelligent anomaly detection and hypothesis generation.

Network Intelligence performs complex pattern analysis on the network data to identify the anomalies across the supported domains. Network Intelligence provide more accurate analysis on anomalies based on the individualized and specific network context.

Functions of Network Intelligence

The primary functions of the Network Intelligence are as follows:

Anomaly detection
Network Intelligence analyzes the Internet Protocol (IP) routers data and detects the anomalies with the machine learning and time-series foundational AI models.
Root-cause hypothesis
Network Intelligence generates the root-cause hypothesis for the anomalies with the time-series foundational models.
Remediation plan
Network Intelligence generates the possible causes and action plan based on the domain knowledge and network context documents to resolve the anomalies in the network operations.
Agentic AI framework
Network Intelligence uses the AI agents to process and match the observation data with the domain knowledge and network context documents to identify the possible causes and remediation plans for the network anomaly.
Service API integration
Network Intelligence supports API integration of services such as information technology service management (ITSM), inventory systems, and data with OpenAPI specification. To integrate service API in the Network Intelligence, see Integrating service APIs in IBM® Network Intelligence.
Automated ticket creation
Network Intelligence creates a ticket with the root-cause hypothesis and applicable remediation automatically for the network anomalies. You need to integrate ITSM in the Network Intelligence to create the ticket automatically for the anomalies.
Model Context Protocol (MCP) server integration
Network Intelligence supports MCP server integration to connect with the external monitoring or management platforms. Network Intelligence receives and analyzes the network data from the MCP servers. To integrate MCP servers in the Network Intelligence, see Integrating Model Context Protocol (MCP) servers.
IBM® watsonx Chat Assistant
The watsonx Chat Assistant is an AI-powered assistant that is built in the Network Intelligence to generate responses for the chat query. Click View reasoning to view the chain-of-thought of the watsonx Chat Assistant and improve the chat responses.
The Reasoning Time feature in the watsonx Chat Assistant consists of Quick, Standard, and Extended options that allows you to configure the time taken for the response generation.
  • Quick: You can set the Quick option for the fastest response generation suitable for simple and straightforward chat queries.

  • Standard: You can set the Standard option for the balanced response generation. The Reasoning Time feature is set to Standard as default.

  • Extended: You can set the Extended option for the deep analysis suitable for the complex and open-ended chat queries. The watsonx Chat Assistant consumes more time to analyze the input and generate a response.

Approval flow in the chat interface
The MCP or API tools that are integrated in the Tool management need your approval to allow the AI assistant to access the tools in the chat interface. The chat interface displays the following options to allow the AI assistant to use the integrated tools:
  • Run once: Click Run once to allow the AI assistant to access the integrated tool for one time only. If the AI assistant needs to access the integrated tools again, the options are prompted in the chat interface.
  • Run for this session: Click Run for this session to allow the AI assistant to access the integrated tool during the complete chat session. The access to the integrated tools is restricted for the AI assistant when you close the chat session.
  • Skip: Click Skip to restrict the tools access for the AI assistant.

AI pipeline in Network Intelligence

Network Intelligence consists of the following foundational models:

  • IBM Time Series Pulse
  • Meta Llama 3.3

Operating Modes of Network Intelligence

Network Intelligence consists of two operating modes: Assistant and Agent modes. For more information, see Operating modes of IBM® Network Intelligence.