Skip to content

7.13 Cisco AI Assistant for Networking

Document Information

Item Details
Organization Abhavtech.com
Version 2.0
Last Updated December 2025

1. Overview

1.1 What is Cisco AI Assistant?

Cisco AI Assistant is a natural language interface integrated into Catalyst Center that allows network administrators to interact with the network using conversational queries instead of navigating complex menus.

┌─────────────────────────────────────────────────────────────────────┐
│                    Cisco AI Assistant Architecture                   │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  User Interface                     Backend Services                │
│  ┌─────────────────────┐           ┌─────────────────────┐         │
│  │   Natural Language  │           │  Cisco AI Cloud     │         │
│  │   Query Input       │──────────►│  (NLP Processing)   │         │
│  │                     │           │                     │         │
│  │  "Show me unhealthy │           │  • Intent parsing   │         │
│  │   devices in Mumbai"│           │  • Context mapping  │         │
│  └─────────────────────┘           │  • Query generation │         │
│           │                        └──────────┬──────────┘         │
│           │                                   │                     │
│           │                                   ▼                     │
│           │                        ┌─────────────────────┐         │
│           │                        │  Catalyst Center    │         │
│           │                        │  API Execution      │         │
│           │                        │                     │         │
│           │                        │  • Device inventory │         │
│           │                        │  • Assurance data   │         │
│           │                        │  • Configuration    │         │
│           ▼                        └──────────┬──────────┘         │
│  ┌─────────────────────┐                      │                    │
│  │   Response Display  │◄─────────────────────┘                    │
│  │                     │                                           │
│  │  "Found 3 unhealthy │   Formatted results with                  │
│  │   devices..."       │   actionable insights                     │
│  └─────────────────────┘                                           │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

1.2 Key Capabilities

Capability Description Example Query
Network Health Query device and client health "What's the health of Mumbai site?"
Troubleshooting Identify and diagnose issues "Why is client AA:BB:CC having issues?"
Inventory Search devices and clients "Show all switches running 17.9"
Configuration Query configurations "Which ports have PoE enabled?"
Assurance Get insights and recommendations "What issues need attention today?"

2. Enabling AI Assistant

2.1 Prerequisites

Prerequisites:

  Catalyst_Center:
    Version: 2.3.5.x or later
    License: DNA Advantage

  Cloud_Connectivity:
    - Outbound HTTPS to Cisco Cloud
    - DNS resolution required
    - Firewall rules for *.cisco.com

  User_Requirements:
    - Admin role or appropriate RBAC
    - Enabled AI Assistant access

2.2 Configuration Steps

Catalyst Center → System → Settings → Cisco AI Analytics

1. Enable Cloud Connectivity
   ☑ Enable Cisco AI Analytics
   ☑ Share data with Cisco Cloud (anonymized)

2. Configure Data Sharing
   Data Types to Share:
     ☑ Network topology
     ☑ Device inventory
     ☑ Assurance metrics
     ☑ Issue data
     ☐ Configuration details (optional)

3. Enable AI Assistant
   ☑ Enable AI Assistant for users

4. User Access Control
   Roles with AI Assistant Access:
     ☑ SUPER-ADMIN-ROLE
     ☑ NETWORK-ADMIN-ROLE
     ☐ OBSERVER-ROLE

5. Save and Apply

2.3 Verify Cloud Connectivity

System → Settings → Cloud Connectivity

Status: Connected ✓
Last Sync: [timestamp]
Data Uploaded: [size]

Connectivity Test:
  ✓ DNS resolution
  ✓ HTTPS connection
  ✓ Certificate validation
  ✓ Authentication

3. Using AI Assistant

3.1 Accessing AI Assistant

Catalyst Center Dashboard → Click AI Assistant icon (top right)

Or use keyboard shortcut: Ctrl + Shift + A

AI Assistant Panel opens on right side of screen

3.2 Query Examples by Category

Network Health Queries

User: "What is the overall network health?"
AI: "The overall network health is 94%. Here's the breakdown:
     - Device Health: 97% (3 devices with issues)
     - Client Health: 92% (125 clients with poor experience)
     - Application Health: 93%
     Click here to see details..."

User: "Show me unhealthy devices"
AI: "Found 3 unhealthy devices:
     1. MUM-ED-05 (Critical - Unreachable since 10:15 AM)
     2. CHE-AP-12 (Warning - High CPU 89%)
     3. LON-ED-08 (Warning - Memory 85%)
     Would you like to troubleshoot any of these?"

User: "What's wrong with Mumbai site?"
AI: "Mumbai site health is 89%. Issues detected:
     - 2 switches with high CPU utilization
     - 15 clients with poor wireless experience
     - 1 AP with channel interference
     Recommended actions: [View Details]"

Client Troubleshooting Queries

User: "Why is client AA:BB:CC:DD:EE:FF having connectivity issues?"
AI: "Analyzing client AA:BB:CC:DD:EE:FF...

     Issues Found:
     1. Poor RSSI (-78 dBm) - Client is far from AP
     2. High roaming (12 roams in last hour)
     3. Authentication delays (avg 2.1 seconds)

     Root Cause: Client is in coverage hole area

     Recommendation: 
     - Add AP coverage in Building A, Floor 2
     - Or move client closer to existing AP

     [View Client 360]"

User: "Show clients with authentication failures today"
AI: "Found 23 clients with authentication failures:
     - 15 failures: Wrong password
     - 5 failures: Certificate expired
     - 3 failures: User not in database

     Top affected users:
     1. jsmith@abhavtech.com (5 failures)
     2. mwilliams@abhavtech.com (3 failures)

     [View ISE Reports]"

Device Queries

User: "List all switches running IOS-XE 17.9"
AI: "Found 45 switches running IOS-XE 17.9.x:
     - 17.9.4a: 30 devices
     - 17.9.3: 10 devices
     - 17.9.2: 5 devices

     Recommendation: 15 devices should upgrade to 17.9.4a
     [View Device List] [Schedule Upgrade]"

User: "Which devices have high CPU?"
AI: "Found 5 devices with CPU > 80%:
     1. MUM-ED-05: 92% (Spanning Tree reconvergence)
     2. CHE-BN-01: 85% (High LISP activity)
     3. LON-ED-12: 83% (ACL processing)
     4. NJ-CP-01: 81% (Map-cache updates)
     5. DAL-ED-03: 80% (PoE negotiations)

     [View Details] [Create Incident]"

Configuration Queries

User: "Show ports with PoE disabled"
AI: "Found 127 ports with PoE disabled:
     - 45 uplink ports (expected)
     - 62 unused ports (expected)
     - 20 access ports (review recommended)

     Access ports with PoE disabled:
     Site: Mumbai - 12 ports
     Site: Chennai - 5 ports
     Site: London - 3 ports

     [View Port List]"

User: "What VLANs are configured on MUM-ED-01?"
AI: "VLANs on MUM-ED-01:
     - VLAN 10: Corporate (42 ports)
     - VLAN 20: Voice (38 ports)
     - VLAN 30: Guest (4 ports)
     - VLAN 40: IoT (8 ports)
     - VLAN 999: Quarantine (0 ports)

     Total: 5 VLANs, 92 assigned ports"

4. AI-Powered Insights

4.1 Proactive Recommendations

AI Assistant proactively surfaces:

┌─────────────────────────────────────────────────────────────────────┐
│                    AI Insights Panel                                 │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  🔴 Critical (2)                                                    │
│  ├── Device MUM-ED-05 unreachable - Immediate action required       │
│  └── Certificate expires in 7 days on ISE-PSN-01                    │
│                                                                     │
│  🟡 Warning (5)                                                     │
│  ├── 3 devices approaching memory threshold (>80%)                  │
│  ├── Wireless interference detected in Building A                   │
│  └── 15 clients experiencing poor DNS resolution                    │
│                                                                     │
│  💡 Optimization (3)                                                │
│  ├── 12 APs could benefit from RRM channel adjustment               │
│  ├── 5 switches have unused PoE capacity                            │
│  └── Network segmentation can be improved in Guest VN               │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

4.2 Trend Analysis

User: "Show network trends for the past month"
AI: "Network Trend Analysis (Last 30 Days):

     📈 Improving:
     - Client onboarding time: 2.8s → 2.1s (-25%)
     - Authentication success rate: 98.5% → 99.2%
     - AP health score: 91% → 95%

     📉 Declining:
     - Switch CPU average: 45% → 52%
     - Wireless client capacity: 78% → 85%

     ⚠️ Attention Needed:
     - Endpoint growth rate suggests capacity planning
       needed in Q2 2026
     - 3 switches approaching end of support

     [View Detailed Report]"

5. Integration with Workflows

5.1 Ticket Creation

User: "Create a ticket for the unreachable device"
AI: "Creating ServiceNow incident...

     Incident Details:
     - Short Description: Device Unreachable - MUM-ED-05
     - Priority: P2
     - Assignment Group: Network Operations
     - Description: [Auto-populated with diagnostic data]

     Incident INC0012345 created successfully.
     [View in ServiceNow]"

5.2 Remediation Actions

User: "Reboot the problematic AP"
AI: "⚠️ Confirmation Required

     Action: Reboot AP CHE-AP-12
     Impact: 23 connected clients will be disconnected

     Are you sure you want to proceed?
     [Confirm] [Cancel]"

User: "Confirm"
AI: "Initiating reboot of CHE-AP-12...
     ✓ Reboot command sent
     ✓ AP is restarting

     Estimated recovery: 3-5 minutes
     I'll notify you when the AP is back online."

6. Best Practices

6.1 Effective Query Patterns

Effective_Queries:

  Be_Specific:
    Good: "Show unhealthy switches in Mumbai"
    Bad: "What's wrong?"

  Include_Context:
    Good: "Why did client jsmith@abhavtech.com fail authentication at 10:15 AM?"
    Bad: "Why did authentication fail?"

  Use_Actionable_Language:
    Good: "List devices that need software upgrade"
    Bad: "Tell me about software"

  Time_Bound_Queries:
    Good: "Show authentication failures in the last 24 hours"
    Bad: "Show authentication failures"

6.2 Query Limitations

What AI Can Do What AI Cannot Do
Query device health Make configuration changes
Search inventory Deploy new configurations
Analyze trends Access external systems directly
Provide recommendations Execute privileged commands
Create tickets (with integration) Bypass RBAC permissions

7. Security and Privacy

7.1 Data Handling

Data_Privacy:

  Data_Sent_to_Cloud:
    - Anonymized topology
    - Device types (not hostnames by default)
    - Aggregated metrics
    - Issue patterns

  Data_NOT_Sent:
    - User credentials
    - Configuration secrets
    - IP addresses (optional)
    - Client PII

  Compliance:
    - Data encrypted in transit (TLS 1.3)
    - Data encrypted at rest
    - GDPR compliant
    - SOC 2 certified

7.2 Audit Logging

All AI Assistant queries are logged:

System → Audit Logs → Filter: AI Assistant

Log Entry:
  Timestamp: 2025-12-28 10:15:23
  User: admin@abhavtech.com
  Query: "Show unhealthy devices in Mumbai"
  Response: "[Device list returned]"
  Duration: 1.2 seconds

8. Abhavtech-Specific Configurations

8.1 Custom Query Templates

Site-specific queries saved as templates:

Template: "Daily Mumbai Check"
Query: "Show health summary for Mumbai site including 
        devices, clients, and top 5 issues"

Template: "Pre-Change Validation"  
Query: "List all active sessions on [device] and 
        show any critical dependencies"

Template: "Post-Incident Review"
Query: "Show all events related to [incident] 
        in the last 24 hours"

8.2 Scheduled Reports via AI

AI can generate scheduled insights:

Daily: "Executive summary of network health"
Weekly: "Top 10 issues and recommendations"
Monthly: "Capacity planning report"

Delivery: Email to noc@abhavtech.com

Document Version: 2.0 Abhavtech.com - SD-Access Implementation