Patents

Machine learning identifies previously resolved alerts that resemble a new incoming alert, finds the properties those resolved alerts share, and — where statistical thresholds are met — has the new alert automatically inherit them.

A natural language request is passed to a machine learning model that identifies the content page matching the user's intent. The page is analyzed on the fly for its interactive elements, and the one corresponding to the request is surfaced to guide the user.

Pre-defined conversation flows are mapped to conversational expressions with matching semantic meaning. When a user's phrasing falls within a similarity threshold of a stored expression, the virtual agent recognizes the topic and runs the corresponding flow — no bespoke training needed.

An AR device captures a real object in view, recognizes it as a managed asset on the network, and sends its identifier plus location and environment context to the management platform. The returned management data is rendered as a virtual object superimposed on the live image.

A conversation service receives a user's natural language description of a task, resolves its intent against a set of conversation trees, performs the action implied by that intent, and selects a response from the action result to return through the virtual agent.

Incidents from a client instance are grouped into clusters by the natural language similarity of a target field. Each cluster is scored for quality and ranked by size, identifying where a service management operation — such as automation or a knowledge article — will have the most impact.