Skip to content

AI Ops🔗

Installation

For information on installing or updating AI Ops, see the section on AI Service System Applications.

AI Ops is an AI-powered messaging assistant designed to help network professionals monitor and configure their network architecture. Through its intuitive interface, the assistant enables administrators to quickly diagnose problems and identify solutions.

Getting Started with AI Ops🔗

Click the AI Ops Panel button to open AI Ops.

Chat Message Interface🔗

The chat message interface is where users interact with AI Ops.

  1. AI Ops Response Field
  2. Search Documents
  3. Build Button
  4. User Submission Field
  5. Help Button
  6. Prompt Submission Button
  7. Name Collision Warnings

Fuzzy Matching🔗

AI Ops is designed to anticipate user intent when object names are queried. Since object names may occasionally differ slightly from what users type, AI Ops employs a technique known as fuzzy matching to interpret and correct user input. For example, if a user searches for an object named "Spine-1" but enters "Spine 1" (omitting the hyphen), AI Ops will attempt to identify and match the intended object. AI Ops notifies the user of the change by posting the word "corrected" beneath the query.

If the current query results in multiple potential matches—creating naming collisions—fuzzy matching is temporarily disabled. In such cases, AI Ops prompts the user to clarify their query.

Help🔗

AI Ops provides a useful help menu with examples of its capabilities. To view Help, click the question mark icon in the upper right corner of the AI Ops chat window.

Note

AI Ops scope is limited to specific objects and tasks. This documentation provides a general overview of its capabilities. For a detailed list of applicable objects and actions please view the Help option in the application.

Behaviors🔗

AI Ops responds to user prompts in the chat message window by displaying data in various formats and visual styles. In addition to displaying data, AI Ops can also respond to user queries by executing site-wide commands that impact overall system operations. This distinction between displaying network information and executing system-altering operations is organized in the documentation under two categories: Data Presentation Behaviors and System Change Behaviors.

Data Presentation Behaviors🔗

Data Presentation Behaviors refer to the ways in which AI Ops presents data to users.

Display Jumps🔗

AI Ops responds to navigation commands by performing a jump and quickly navigating to the referenced object. A command such as Navigate to Leaf 7 will center the view window to Leaf 7, making the object the focal point.

Displaying Objects🔗

AI Ops displays objects that match the query scope. The query scope can include a single object or many objects.

In the example below, a device named Leaf7 is returned from the query.

Object Information🔗

Objects are displayed with information about their status. Each information symbol is described in the diagram below.

  1. Status indicator
  2. Provisioning indicator
  3. Alarm indicator (Alarms are notifications that inform the administrator about changes in the network state)
  4. Object Button (The Object button behaves the same as it does throughout the rest of the application. Hover, left click and right click functionality is identical)
  5. Name
  6. Context (Describes the containing object or environment)

Show/Hide Lists🔗

Lists have a Hide/Show feature that allows users to collapse (hide) and expand (show) lists in the chat.

Verity Documentation Reference🔗

When prompting AI Ops, material from the official documentation may be presented and/or requested by the user. This information is usually presented in aggregate form from different sections of the documentation, with individual reference links to the source material included. Responses may include documentation content as part of a larger reply, or users can directly request official documentation by enabling the Docs (Search Documents) button .

Reports🔗

Report data is presented directly in the chat window with hyperlinks to the source Report(s).

Highlighting🔗

Highlighter is a powerful tool for tracing an object usage anywhere in the UI. When prompting AI Ops to use the highlighter feature, the selected object and any trace objects associated with it are highlighted in purple.

The instant search allows users to quickly find objects by name. To use this feature, preface the search with the @ symbol and type the name of the object or device you are searching for. Use the up and down arrows to choose from the presented items, and press Tab to auto-populate the selection in the submission field.

Instant Search for Statistics🔗

Instant search lets you quickly view different statistical descriptors for devices and provisioning objects using this format: @stat:{stat name}.

In this example, all switches with a CPU usage of less than 80 percent are listed.

System Change Behaviors🔗

System Change Behaviors are actions that change the system state. There are two forms of System Change Behaviors:

  • Device Operations: These are system changes made using AI Ops that do not require Build mode to be enabled.
  • Provisioning Actions: These are CRUD (create, read, update, and delete) actions applied to provisioning objects that require Build mode to be enabled.

Device Operations🔗

The current version of AI Ops supports the following Device Operations.

Confirmation Dialog🔗

All Device Operations return a validation message that users must approve in order to commit the change(s).

Load Network Time Traveler Backup🔗

Prompting "list time travelers" lists Time Traveler instances (). Click the restore button to revert to a previous state by loading a selected instance.

A validation message is presented asking the user to apply the changes or cancel.

Provisioning Actions🔗

To create, remove, update or delete provisioning objects the Build button must be enabled. Prompt commands that require the Build option to be enabled are called Provisioning Actions.

To commit Provisioning Actions you must submit the text "Apply changes".

Naming Collision Warnings🔗

Provisioning objects with duplicate names (using the same name more than once), name collisions (names that are too similar), or common words (such as naming objects with generic terms like 'a') will trigger a notification icon with a red dot at the top of the AI Ops messaging window. Clicking this icon opens a notification window that describes the naming issue.