segment-4-agentic-orchestration

Segment 4: Agentic Orchestration & Resilience

Duration: 50 minutes Level: Advanced Last Updated: June 2026

Warner's Laws Covered

Law	Title	Core Idea
23	LLM Personality Matching	Different models have different strengths; match the model to the task
24	Subagent Orchestration	Parallel subagents dramatically reduce cycle time for complex workflows
25	Checkpoint Before Consequence	Always create a rewind point before any high-risk operation
26	Expect Breaking Changes	Agentic tooling evolves fast; build for adaptability, not permanence

Scenario: Contoso Robotics

All exercises use a shared fictional company:

• Company: Contoso Robotics
• Size: 500 employees, $120M annual revenue
• HQ: Austin, TX
• CEO: Maria Chen
• Products: WarehouseBot Pro, WarehouseBot Lite, LogiMover 500
• Strategic Goal: Evaluate entry into the healthcare logistics robotics market

File Inventory

Demos

File	Law	Description
demos/llm-personality-test.md	23	Send the same Contoso prompt to ChatGPT, Claude, and Gemini; score outputs with a rubric
demos/subagent-orchestration.md	24	Parallel subagent walkthrough in Claude Code: security review, test generation, and doc updates
demos/checkpoint-rewind-exercise.md	25	Risky refactor with checkpoint and rewind in Claude Code
demos/copilot-studio-agent.md	26	Build a Contoso customer support agent in Microsoft Copilot Studio

Data

File	Purpose
data/contoso-llm-comparison-rubric.csv	Scoring rubric CSV for the LLM personality test
data/contoso-agent-knowledge-base.md	Knowledge base document for Copilot Studio upload
data/contoso-healthcare-market-brief.md	Grounding document for the healthcare logistics prompt

Live Demo

Folder	Description
mcp-demos/	Working MCP weather server used for the live Model Context Protocol demo (see its own README for setup)

Prerequisites

• Claude Pro or API access (for Claude Code exercises)
• ChatGPT Plus or Team account
• Google Gemini Advanced
• Microsoft 365 with Copilot Studio license (for the Copilot Studio exercise)
• Git installed locally

Quick Start

1. Read data/contoso-healthcare-market-brief.md to understand the scenario context.
2. Open the demo files in order (Laws 23 through 26).
3. Use data/contoso-llm-comparison-rubric.csv to score outputs from the LLM personality test.
4. Upload data/contoso-agent-knowledge-base.md as a knowledge source in Copilot Studio.