Agentic AI Discovery Lab
This common lab moves participants from agent awareness to hands-on delivery. Every learner builds three working telco-oriented agents, uses real model routing for text, vision, and audio, works with reusable skills, and executes multiple actions through guided workflows without needing custom integration.
.svg){kind=small}
About The Lab
Prerequisites
Audiences
Lab Architecture
The Lab architecture seamlessly integrates multiple AI models, agentic workflows, and MCP connectors to create a comprehensive environment for AI-driven network operations, which includes a multi-model layer (Claude Haiku 4.5 as default, GPT-5.2 via codex exec, Pixtral-large for vision) and three specialized agent editions (Runbook Assistant, Field Edition, and Ops Edition with free5gc and MCPs) that allow the user to test different end-to-end use cases.
Why this Lab ?
This lab moves participants from agent awareness to hands-on delivery. Every learner builds three working telco-oriented agents, uses real model routing for text, vision, and audio, works with reusable skills, and executes multiple actions through guided workflows without needing custom integration.
Lab Objectives
- Understand the building blocks of an AI agent: role, instructions, actions, approval gates, skills, and MCP connectivity.
- Route routine work to low-cost real models and reserve deeper reasoning models for the right checkpoints.
- Create agents that can work across text, PDF, image, and audio evidence in realistic telecom scenarios.
- Use reusable skills to ground answers in source material, normalize field evidence, and generate operator handoffs.
- Query sanitized telco systems through mock MCP tools and resources and use the responses in agent flows.
