Episode 1 - Child Agents vs Connected Agents | In Our AI Era with Ragnar and Tina
One agent trying to do everything is where most AI projects fail. We built the framework to fix that. This is Episode 1 of In Our AI Era with Ragnar Pitla and Tina van Heerden. Two practitioners from Microsoft's Agentic team, breaking down what we're actually seeing and building in the AI space. Real projects, real customers, no hype. In this episode, we cover the two multi-agent patterns you need to know in Copilot Studio: Connected Agents and Child Agents. We walk through when to use each, show two real architectures (customer service + Order to Cash ERP), and give you a decision framework you can apply this week. What you'll learn: Why single agents break as your use case grows Connected agents: independent agents that hand off across system boundaries Child agents: specialized subtasks within the same workflow How MCP (Model Context Protocol) made multi-agent practical overnight Practical tips from the field: descriptions, agent count, republishing, conversation history Timestamps: 0:00 The question we get every week 1:00 Welcome to In Our AI Era 2:00 Tina's snowshoeing analogy (this one lands) 4:00 Why single agents hit a wall 7:00 Connected agents explained 9:30 Child agents explained 15:00 Customer service walkthrough (espresso machine example) 19:00 ERP walkthrough (Order to Cash with D365) 25:00 Why MCP changed everything 28:00 Decision framework: connected vs child 29:00 Tips from the field 34:00 Your challenge this week Links: Copilot Studio multi-agent training (build a US + UK child agent): https://microsoft.github.io/mcs-labs/... Writing Agent Instructions and Descriptions: https://learn.microsoft.com/en-us/mic... Decision framework table: https://inouraiera.com/ Your challenge: Open Copilot Studio, find one agent where you're cramming too much in, and split it. Drop a comment describing your scenario and we'll tell you which pattern fits. New episodes regularly. Subscribe so you don't miss the next one! Views expressed are our own and do not represent Microsoft's official position.

Using Large Language Models | Build Your Own LLM Workshop #1

Hermes Architecture EXPLAINED: Memory, Context & Gateways

A2A vs MCP: AI Agent Communication Explained

Episode 2 - Are skills replacing prompts? | In Our AI Era with Ragnar and Tina

How AI agents & Claude skills work (Clearly Explained)

Inside Anthropic, the $965 Billion AI Juggernaut | The Circuit

Episode 8 - The most interesting announcements at Microsoft Build 2026 | In Our AI Era

Billionaire's WARNING: I'm SELLING. The Crash Is Already Here!

What AI Agent Skills Are and How They Work

Turing Award Winner: Disagreeing with Google, Postgres, Future Problems | Mike Stonebraker

Model Context Protocol (MCP) Explained for Beginners: AI Flight Booking Demo!

MCP vs API: Simplifying AI Agent Integration with External Data

How to Build & Sell AI Agents: Ultimate Beginner’s Guide

Episode 7 - The age of Autopilots: Microsoft Scout vs Copilot Cowork | In Our AI Era

The Future of AI Agents with Andrew Ng | Interrupt 26

This Month at Salesforce | Episode 1 – Agentforce, Headless 360, Slack

Skill Issue: Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI

From Idea to $650M Exit: Lessons in Building AI Startups

Episode 4 - Hosted Agents | In Our AI Era with Ragnar and Tina

