Sessions at Cisco Live EMEA Amsterdam demonstrating how to build agentic AI-driven network operations using Cisco Workflows, ThousandEyes, and CX RADKit.

Authors

• Scott Dozier - Lab Developer
• Steve Holl - Lab Developer
• Aman Sardana - Contributor

BRKOPS-2303: Ambient AI Agents Running your network while you sleep

A 1-hr technical breakout deep diving into what ambient ambient agents are and demonstrating their application. Looking to build or replicate what you saw in this session? Refer to Lab 3 and Lab 4.

LTRAI-1487: Cisco Workflows AI Agents Lab

This is a 4 hour lab where you get hands on experience building autonomous event-driven response, starting with static workflows and building up to observability-driven agentic closed loop response.

Lab Guide

The full lab guide is available online at:

🌐 https://cl-ltr.ciscolabs.com/0361f55971/

You can also serve the guide locally with MkDocs:

cd lab-guide
pip install mkdocs
mkdocs serve

Overview

This lab guides participants through building an end-to-end agentic network operations pipeline—from event detection through AI-driven troubleshooting and automated remediation. You'll progress from basic notification workflows to fully autonomous AI agents that can diagnose and resolve network issues.

Here is a 10m video demo of what you will build.

Architecture

Device Events → Splunk/ThousandEyes → Cisco Workflows → AI Agent → RADKit → Device Remediation

Labs

Lab 1: Setting up Network Topology

Configure the foundational infrastructure for agentic network operations:

• Access dCloud lab environment
• Configure router syslog forwarding to Splunk
• Set up Splunk webhooks to trigger Cisco Workflows
• Build a basic notification workflow with Webex Teams integration

Lab 2: Automated Response with Cisco Workflows

Build rule-based remediation workflows for closed-loop automation:

• Register a Cisco Workflows Remote Server for device connectivity
• Configure terminal targets
• Create automated remediation workflows (interface unshut)
• Parse webhook payloads and dynamically select target devices

Lab 3: Agentic AI with RADKit Integration

Import and configure the AI agent for cognitive troubleshooting:

• Import the agentic AI workflow into Cisco Workflows
• Install and configure RADKit with MCP server integration
• Connect the AI agent to interface shutdown events
• Observe AI-driven analysis and remediation recommendations

Lab 4: ThousandEyes Integration for Digital Experience Monitoring

Extend agentic response with ThousandEyes observability:

• Register and configure a ThousandEyes Enterprise Agent
• Set up HTTP Server tests and alert rules
• Integrate ThousandEyes webhooks with Cisco Workflows
• Trigger AI-driven troubleshooting from network performance degradation
• Observe end-to-end root cause analysis and automated remediation

Lab 5: (Optional) Integrate your own tool

Learn how to build and intergrate your own tools/integtations to work with the AI agent

Repository Structure

├── cfg/                    # Router configurations (R1, R2, R3)
├── lab-guide/              # MkDocs-based lab documentation
│   └── docs/               # Lab markdown files
├── scripts/                # Setup scripts
│   └── mcp-server/         # MCP server enrollment and setup
│   └── tools/              # Script to update JSON for AI agent tool list
│   └── radkit/             # Radkit setup--not needed for the lab
└── workflows/              # Cisco Workflows JSON exports

Prerequisites

• Access to Cisco dCloud lab environment
• Cisco Workflows account (meraki.cisco.com)
• Webex account for bot integration
• ThousandEyes account (for Lab 4)

Supporting info

• Cisco Workflows
• Cisco CX RADKit
• Unofficial RADKit Community MCP server - Note: an official MCP server is coming soon for RADKit
• Unofficial Workflows/XDR remote server registration script

License

This project is intended for educational purposes as part of Cisco Live EMEA 2026.