Network Intelligence AI

Create Source

A source is required to connect DvSum to your network data system such as NXT. This connection enables data ingestion, which is essential for generating incidents and running AI analysis.

• Go to Data Sources in Administration 
• Click Add Source
• Select a data source type (MySQL for NXT)
• Select source type (Cloud / On-Prem) 
• Enter all the required connection details.
• Click Test Connection and Save

Default Configuration

Once the NXT source is linked, the system automatically provides the required setup, including:

• Default Collection
• Pre-configured Endpoints (Queries)
• Default Network Analysis Workflow

These components are pre-configured and ready to use, so manual creation is not required.

Note:
Users can review or modify these configurations if needed, but for standard usage, the default setup is sufficient.

Execute Workflow

The workflow is executed to process the ingested data and generate Fiber Node Incidents.
This step triggers the AI analysis using the ingested network data.

Steps:

1. Open the Network Analysis Workflow
2. Click Execute Now

Alternative Option

• Click on the three dots (⋮)
• Select Add to Job to schedule execution

Notes

• Execute Now runs the workflow immediately without creating a job
• Use Add to Job for scheduled or managed execution
• Execution can be monitored from:
  • Jobs
  • Job Execution

Cloning Workflow

The Proactive Workflow can be cloned by selecting Clone from the three-dot (⋮) menu. The cloned workflow is created as a separate asset, allowing its all details to be modified independently of the original collection.

Incident Analysis

After workflow execution, Fiber Node Incidents are automatically created based on the configured queries and ingested data.

1. Navigate to Asset Dictionary
2. Select Fiber Node Incident as the asset type
3. Open the required Incident record
4. Go to the Analysis tab

Analysis Page Sections

Section	Details	Purpose
Operational Snapshot	- Primary Driver (e.g., USFEC)  - Driver Details  - Impacted Devices  - Impact Radius  - Time Behavior  - Priority Age	Provides a quick overview of the issue and its impact scope
RCA & Intelligence	- Suspected Root Cause  - Confidence Level  - Suggested Actions  - Evidence	Explains why the issue occurred and suggests next steps
Summary	- Status  - Impact %  - Pattern  - Persistence	Gives a high-level business view of the incident
History / Trend Graph	- Priority trend (P0, P1, etc.)  - Metrics like USFEC, Connectivity	Helps analyze behavior and trends over time
Cable Modems Table	- Primary Issue  - Service Impacted  - Status  - Reset Count	Provides device-level details for deeper investigation

Actions on Incidents

Users can take various actions on a Fiber Node Incident directly from the Actions menu in the incident detail page.

Update Status

Acknowledged

Marks the incident as acknowledged, indicating that it has been reviewed and is being worked on.

Ticketed

Indicates that a ticket has been created for this incident for tracking.

Resolved

Marks the incident as resolved after the issue has been fixed.

Closed

Closes the incident after resolution and validation.

Ignored

Marks the incident as not relevant or not requiring further action.

Route Actions

Re-assign To

Allows users to assign the incident to another user for further investigation or action, with a notification sent via email.

More Actions

Suppress

Used when the incident is known or does not require attention for a certain period.

Suppresses the incident for a specified duration without changing its status. Users are required to enter the suppression time period. During this period, no email notifications will be sent for any updates related to the incident.

Example Use Case

If a user identifies an incident caused by a planned maintenance activity on a fiber node.

Since the issue is expected and already being handled, the user suppresses the incident for a specific duration (e.g., 4 hours) to avoid unnecessary alerts.

This ensures that:

• No email notifications are triggered during the maintenance window
• Normal monitoring resumes automatically after the suppression period ends

Submit AI Feedback

Allows users to provide feedback on AI-generated insights, helping improve future analysis accuracy.

Activity 

The Activity section provides a detailed log of all actions and updates performed on a Fiber Node Incident. It helps users track the complete history of the incident lifecycle.

Watch Incident

The Watch option allows users to subscribe to specific incident updates by selecting the conditions under which they want to receive alerts.

Example Use Case

A network operations engineer is monitoring a fiber node incident that is currently low priority (P3) but could impact customers if it worsens.

Instead of tracking the incident manually, the user configures Watch to:

• Get notified when priority increases (e.g., P3 → P1)

This ensures the user is alerted only when:

• The issue becomes critical and requires immediate attention

Priority Increases- Receive notifications when the incident priority is escalated (e.g., P2 → P1).
Remains at current priority for configured number of hours- Receive alerts if the incident stays at the same priority for a configured number of hours.

How It Works

• Users can select one or multiple conditions
• Notifications are triggered only when the selected conditions are met

Re-Analyze Node

Used when users want to refresh analysis after new data is available or to re-evaluate the issue.

Triggers the system to rerun analysis on the selected node in online mode, immediately refreshing insights, root cause, and recommendations without requiring scheduling

For configuration steps and instructions on using the Aura AI Agent, refer to the article Configure and Use the Aura AI Agent.

Configure Aura AI Agent

The Aura AI Agent provides a simplified configuration experience by allowing users to associate the agent with a Proactive Workflow.

Steps

1. Navigate to Data Sources → {REST API Source Name} → AI Agents.
2. Click Add Agent.
3. Select Aura AI Agent.
4. Under Workflow Assignment:
   • Select the required Solution Type (for example, HFC Fiber Node or HFC Driver).
   • Select the required Proactive Workflow.
   • Select the Reasoning Effort:
     • Fast – Prioritizes faster responses.
     • Balanced – Balances response speed and analysis quality.
     • Deep – Performs more comprehensive analysis.
5. Click Save.

Notes

• The available Solution Types are configured under Administration → Account →  Module Settings → AI Orchestration.

  

• The selected workflow is associated with the Aura AI Agent and can be viewed from the workflow details.
• Each Aura AI Agent can be associated with only one Proactive Workflow.

Once the agent is saved, the selected Proactive Workflow is automatically associated with the agent. You can verify this association from the corresponding Workflow details.

Interact with the Aura AI Agent

After the Aura AI Agent is configured, its behavior can be validated using Train Agent or Talk to Your Data before executing the Network Analysis Workflow.

1. Navigate to the configured Aura AI Agent.

2. Open Train Agent.

   

3. Enter a prompt related to the Analyze Fiber Node workflow.
4. Review the AI-generated response.

Using Talk to Your Data

The Talk to Your Data page enables users to interact with the Aura AI Agent and access conversations related to analyzed Fiber Node Incidents.

When a Fiber Node Incident is analyzed from its Engage tab, a conversation is automatically created and listed in Talk to Your Data using the corresponding Fiber Node ID.

Selecting the conversation opens the associated Fiber Node Incident, allowing users to continue reviewing the same analysis context.

Conversation history is retained and available from the conversation list.

History

The History page allows users to view and manage previous conversations with AI Agents. It provides quick access to ongoing and past conversations, along with details such as the associated AI Agent, source asset, and recent activity.

The page contains the following tabs:

All Chats

The All Chats tab displays all conversations created with AI Agents.

For each conversation, the following information is available:

• Chat Name – Name of the conversation.
• AI Agent – AI Agent associated with the conversation.
• Questions Asked – Number of questions asked during the conversation.
• Source Asset – Asset associated with the conversation, such as a Fiber Node Incident.
• Last Activity – Date and time of the most recent interaction.
• Created By – User who created the conversation.

Users can search, filter, sort, and paginate through the conversation list.

Bookmarked Chats

The Bookmarked Chats tab displays conversations that have been bookmarked for quick access, allowing users to easily revisit important discussions.

Users can bookmark conversations in either of the following ways:

• From the History page by selecting one or more conversations and choosing the Bookmark action.

  

• From within an individual conversation by selecting the Bookmark option.

  

Bookmarked Questions

The Bookmarked Questions tab displays bookmarked questions from conversations. Users can quickly access frequently referenced questions and their corresponding responses without searching through the entire conversation.

Users can bookmark questions from within an individual conversation by selecting the Bookmark option for the required question.

Archived Chats

The Archived Chats tab displays conversations that have been archived. Archived conversations remain available for future reference while being separated from active conversations.

Users can archive conversations in either of the following ways:

• From the History page by selecting one or more conversations and choosing the Archive action.

  

• From within an individual conversation by selecting the Archive option.