Uploaded September 2026 | Updated November 2026, 7 hours ago
Every application contains business rules, architectural patterns, internal frameworks, and development standards that generic AI cannot fully understand.
Without that context, modernization efforts can produce inconsistent results and require significant manual review.
Learn how GitHub Copilot Modernization custom skills help organizations bring their institutional knowledge directly into the modernization process.
See how custom skills can encode business logic, software factory patterns, and governance requirements to generate more consistent, accurate, and production-ready outcomes across applications and teams. Through a live demonstration, you'll see how custom skills transform modernization workflows by turning tribal knowledge into reusable guidance that scales across projects.
Join us to learn how to preserve organizational expertise while accelerating application modernization with AI.
Key Takeaway: Custom skills help organizations modernize faster while ensuring AI-generated changes align with their unique standards, practices, and business requirements.
0:00:02 - Welcome and Introduction
0:00:54 - Modernization Skills and Tribal Knowledge
0:02:43 - Five Key Takeaways
0:04:15 - Demo 1: Modernizing Libraries (Newtonsoft to System.Text.Json)
0:11:15 - Demo 2: Codifying Tribal Knowledge into Skills
0:15:10 - Understanding the Skill Definition Structure
0:21:30 - Demo 3: Testing and Refining Skills
0:31:08 - Demo 4: Troubleshooting and Fixing Regression Bugs
0:40:20 - Demo 5: Reusing Skills for New Features
0:47:15 - Behind the Scenes: Managing Skill Workflows
0:50:40 - Recap: Tools for Modernization (Prompts, Skills, Tests)
0:51:37 - Next Steps: Building a Team Practice
0:53:00 - Resources and Official Documentation
[eventID:27594]
Every application contains business rules, architectural patterns, internal frameworks, and development standards that generic AI cannot fully understand.
Without that context, modernization efforts can produce inconsistent results and require significant manual review.
Learn how GitHub Copilot Modernization custom skills help organizations bring their institutional knowledge directly into the modernization process.
See how custom skills can encode business logic, software factory patterns, and governance requirements to generate more consistent, accurate, and production-ready outcomes across applications and teams. Through a live demonstration, you'll see how custom skills transform modernization workflows by turning tribal knowledge into reusable guidance that scales across projects.
Join us to learn how to preserve organizational expertise while accelerating application modernization with AI.
Key Takeaway: Custom skills help organizations modernize faster while ensuring AI-generated changes align with their unique standards, practices, and business requirements.
0:00:02 - Welcome and Introduction
0:00:54 - Modernization Skills and Tribal Knowledge
0:02:43 - Five Key Takeaways
0:04:15 - Demo 1: Modernizing Libraries (Newtonsoft to System.Text.Json)
0:11:15 - Demo 2: Codifying Tribal Knowledge into Skills
0:15:10 - Understanding the Skill Definition Structure
0:21:30 - Demo 3: Testing and Refining Skills
0:31:08 - Demo 4: Troubleshooting and Fixing Regression Bugs
0:40:20 - Demo 5: Reusing Skills for New Features
0:47:15 - Behind the Scenes: Managing Skill Workflows
0:50:40 - Recap: Tools for Modernization (Prompts, Skills, Tests)
0:51:37 - Next Steps: Building a Team Practice
0:53:00 - Resources and Official Documentation
[eventID:27594]
![Model Mondays - Voice Agents Upgraded: New Speech Models and Agent Experience In Foundry
Speech models let applications understand and generate spoken language, powering experiences like transcription, real-time voice interaction, and accessible conversational interfaces.
In this episode, well walk through a live demo in the Foundry playground that shows how to build, test, monitor, and optimize a voice agent from start to finish. Well also cover the latest development in speech-to-text, text-to-speech, and speech-to-speech models in Azure Speech in Microsoft Foundry, including new MAI models, and how they make voice agents more natural and responsive.
Repo - https://aka.ms/model-mondays
Resources: https://aka.ms/model-mondays/collection
0:00:00 - Introduction and Code of Conduct
0:00:54 - Model Mondays: Welcome and Overview
0:01:54 - Weekly News Highlights: New Models and Infrastructure
0:07:16 - Voice Agents in Microsoft Foundry: Overview and Capabilities
0:14:14 - Live Demo: Building a Voice Agent for Finance
0:34:07 - Audio Models in Microsoft AI: Transcription and Speech
0:44:48 - Speech Models in Foundry: Architectures and Pipelines
0:51:09 - Customer Use Cases and Implementation Q&A
1:00:32 - Wrap-up and Next Episode Announcement
- Explore the Resources
https://github.com/microsoft/model-mondays
- Read the Newsletter https://aka.ms/model-mondays/newsletter
- Continue the conversation on the Discord https://aka.ms/model-mondays/discord
[eventID: 27565] Model Mondays - Voice Agents Upgraded: New Speech Models and Agent Experience In Foundry](https://i.ytimg.com/vi/rBFQLt3StVI/mqdefault.jpg)
![Talvez você já esteja usando mais IA do que imagina
A IA já está aparecendo nas ferramentas que profissionais de dados usam todos os dias, às vezes de formas que nem percebemos.
Vamos explorar o que talvez já esteja disponível para você, onde a IA pode ajudar e onde o seu próprio julgamento continua sendo essencial.
Para mais informações sobre o Data Days e tudo o que ele tem a oferecer, acesse https://aka.ms/DataDays
📌 Este evento faz parte de uma série. Saiba mais aqui: https://aka.ms/AIatWork/UnderstandIt/y
#MicrosoftReactor #LearnConnectBuild #Fabric #AI #DataDays
[eventID:27729] Talvez você já esteja usando mais IA do que imagina](https://i.ytimg.com/vi/rFQJN53piws/mqdefault.jpg)
![The Agent Harness: From GitHub Copilot to Production AI
What if the secret to building better AI agents isnt a bigger model, but a better harness?
In this episode of New Breakpoint, Graeme Foster explores the GitHub Copilot SDK and the emerging agent harness pattern thats powering a new generation of AI experiences.
Discover how planning, tool use, code execution and reasoning come together to help agents solve real-world problems and learn how these capabilities can be extended beyond the IDE into hosted, governed and enterprise-ready solutions.
Key takeaways:
Understand the agent harness pattern and why modern AI systems are about more than just models, combining tools, planning, execution and orchestration to achieve complex goals.
Explore the GitHub Copilot SDK in action, including how agents can analyse data, execute code, interact with files and leverage external tools to complete end-to-end workflows.
Learn how to take agents into production with hosted experiences, identity, governance and integration with Foundry, enabling scalable and enterprise-ready AI solutions.
0:04 Introduction to New Breakpoint
0:58 Meet Graeme Foster and the Agent Harness
5:07 Redefining Complex Agentic Systems
7:33 What Is an Agent Harness?
15:05 GitHub Copilot SDK and Agent Examples
22:10 The .NET Development Experience
26:45 Using Skills for Progressive Disclosure
32:09 Data Analysis with the Claims Tool
40:46 Deploying to Microsoft Foundry
48:33 Identity and Governance for Agents
52:05 Closing Thoughts and Future Outlook
[eventID:27595] The Agent Harness: From GitHub Copilot to Production AI](https://i.ytimg.com/vi/rTGX69aa-z8/mqdefault.jpg)
![Episodio 2: IA Generativa – Creando, Innovando y Acelerando con IA
Explora cómo la IA Generativa puede crear contenido, generar ideas, resolver problemas y aumentar la productividad. Los participantes aprenderán cómo la IA está transformando la creatividad, la colaboración y la innovación en diversos sectores.
🔗 Aprende más accediendo a los recursos: https://aka.ms/conceptosdeIA
📌 Este evento es parte de una serie, aprende más aquí: https://aka.ms/aiskillsfortomorrow/es/y
00:00:00 Introducción y bienvenida
00:02:10 Conceptos de IA Generativa, prompts y contexto
00:03:54 Introducción a Microsoft Copilot
00:04:25 Diferencias entre Copilot y Agentes especializados
00:06:21 Demostración: Uso de Copilot en modo trabajo y web
00:11:15 Plataformas para agentes: Microsoft 365 vs. Foundry
00:15:02 Agentes predefinidos: Investigador y Analista
00:23:00 Demostración: Análisis de datos con el Agente Analista
00:27:00 Demostración: Investigación robusta con el Agente Investigador
00:29:22 Los 6 elementos clave para trabajar con IA Generativa
00:33:30 Creación de agentes personalizados en M365 Copilot
00:36:15 Introducción a Copilot Studio y automatización de acciones
00:40:35 Introducción a Azure AI Foundry para desarrolladores
00:43:40 Servicios de IA en Foundry (traducción, speech, OCR)
00:54:40 Integración final y conclusiones de la sesión
#MicrosoftReactor #LearnConnectBuild #MicrosoftCopilot #Copilot #Agents #GenAI
[eventID:27648] Episodio 2: IA Generativa – Creando, Innovando y Acelerando con IA](https://i.ytimg.com/vi/rV7ddXDFt8M/mqdefault.jpg)
![Build your AI Teammate in Teams
Ready to bring an AI teammate into Microsoft Teams? In this hands-on session, well go from zero to a working agent, step by step.
Well start by setting up a starter AI teammate and connecting the services it needs. Next, well use the Teams CLI to register the app and run the teammate on your own machine. To finish, well install it in Teams and test its first conversation, so you leave with a working agent you can keep building on.
📌 This event is part of a series. Learn more here:
https://aka.ms/BeyondTheAgentSeries/y
#MicrosoftReactor #LearnConnectBuild #MicrosoftTeams #TeamsSDK #TeamsCLI #TeamsDeveloper #Agents
[eventID:27791] Build your AI Teammate in Teams](https://i.ytimg.com/vi/rfxm-FxZTAY/mqdefault.jpg)
![Governance at Scale: Enforcing Network Security Baselines with Azure Policy & Landing Zones
This session makes the case for guardrails, not gates: governance thats built into the platform itself rather than bolted on afterward.
Well start with Azure Policy fundamentals, including policy definitions, initiatives, and effects (Deny, Audit, DeployIfNotExists, Modify), and explore how each effect type is appropriate for different governance scenarios.
From there, we move into practical network security baselines, covering policies that:
- Deny public IP creation on VMs
- Enforce mandatory NSG association on subnets
- Require specific Azure Firewall routing
- Mandate encryption in transit
The session then zooms out to the Cloud Adoption Framework Landing Zone model, showing how these individual policies are assembled into a coherent governance architecture and applied automatically at subscription vending time. This ensures every new subscription inherits the right guardrails from day one, rather than having security retrofitted later.
Well also cover:
- Management group hierarchy design
- Policy assignment scope
- Handling policy exemptions for legitimate edge cases without undermining the overall baseline
00:00 Introduction and Session Overview
00:53 Secure Future Initiative and Azure Policy Fundamentals
03:42 Azure Policy Architecture: Definitions, Initiatives, and Assignments
07:42 Built-in Policies and Common Governance Scenarios
08:36 Policy Initiatives and Assignment Scopes
10:58 Understanding Policy Effects: Deny, Audit, Modify, and Remediation
14:02 Azure Policy Best Practices and Decision Guidance
15:08 Introduction to Azure Landing Zones
16:25 Policy Inheritance and Management Group Hierarchy
17:52 Governance at Scale with Landing Zones and Azure Policy
19:04 Network Security Baselines with Azure Policy
20:17 Boundary Enforcement and Private Network Controls
20:56 Virtual Machine and NSG Hardening
21:37 Network Visibility, Logging, and Monitoring Requirements
22:06 Firewall, WAF, and DDoS Protection Policies
23:22 Securing VPN and Connectivity Infrastructure
24:00 Applying Security Benchmarks and Regulatory Frameworks
24:40 Azure Policy Demonstration Overview
25:18 Assigning Policies at the Management Group Level
28:10 Enforcing Regional Compliance with Allowed Locations Policies
29:20 Building Policy Initiatives and Remediation Tasks
31:27 Live Demo: Policy Enforcement Blocking Public IP Assignment
33:40 Firewall Enforcement and Compliance Monitoring
34:35 Key Takeaways and Governance Strategy Recap
35:05 Q&A, Resources, and Closing Remarks
[eventID:27526] Governance at Scale: Enforcing Network Security Baselines with Azure Policy & Landing Zones](https://i.ytimg.com/vi/rpK3ryHcT5I/mqdefault.jpg)
![Episodio 4: Agentes de IA – De la Asistencia a la Colaboración Autónoma
Explora la próxima evolución de la IA a través de agentes inteligentes capaces de razonar, planificar, automatizar y colaborar con los usuarios. Los participantes conocerán casos de uso emergentes y oportunidades para aplicar agentes de IA en los entornos de trabajo modernos.
🔗 Aprende más accediendo a los recursos: https://aka.ms/conceptosdeIA
📌 Este evento es parte de una serie, aprende más aquí: https://aka.ms/aiskillsfortomorrow/es/y
0:00 Introducción y bienvenida
3:14 Mapa de la sesión: Delegar, supervisar y escalar
4:22 Cambio de paradigma: De asistentes a agentes autónomos
5:10 El ciclo de 4 pasos del agente (Percibir, planificar, actuar, observar)
8:43 El contrato operativo del agente
14:10 Cuatro niveles de autonomía en agentes de IA
19:28 Ciclo interno del agente: Observar y replanificar
21:20 Seis preguntas clave antes de dar autonomía a un agente
25:03 Demostración práctica: Construyendo un agente en Copilot Studio
32:05 Componentes del agente (LLM, instrucciones, conocimiento y herramientas)
36:03 Configuración de workflows para autonomía
55:14 Cierre: Delegar el trabajo, conservar el criterio
#MicrosoftReactor #LearnConnectBuild #MicrosoftCopilot #Copilot #Agents #GenAI
[eventID:27650] Episodio 4: Agentes de IA – De la Asistencia a la Colaboración Autónoma](https://i.ytimg.com/vi/rs0dptRRyfk/mqdefault.jpg)
![Composing Knowledge Bases That Reason Over Work, Business, and the Web
Learn how to design knowledge bases that span multiple domains. This session focuses on retrieval strategies that bring together work data, structured business context, and real-time web signals.
We’ll explore how to orchestrate these sources so agents can reason across them effectively, rather than treating each in isolation.
0:00:00 Introduction and Housekeeping
0:00:58 Overview of Foundry IQ and Knowledge Bases
0:03:43 Understanding Knowledge Base Concepts and Sources
0:05:25 End-to-End Conceptual View of Agentic Retrieval
0:08:21 Retrieval Reasoning Effort Levels
0:09:53 Iterative Planning and Review Loop
0:11:18 Iterative Context Engineering
0:12:42 Iterative Search Strategy Example
0:14:00 Output Components: Logs, Merged Results, and Synthesis
0:15:16 Effort Limits for Retrieval Iterations
0:16:19 Data Grounding Options
0:22:06 Knowledge Source Selection Process
0:25:02 Deep Dive into Knowledge Source Types (Blob, SQL, Files, SharePoint)
0:28:08 Advanced Connectors: Work IQ, Fabric, and MCP
0:30:17 Security and Sensitivity Labels
0:31:07 Live Demo: Building a Knowledge Base in Foundry
0:46:05 Code Walkthrough and Python Samples
0:49:57 Q&A Session
🔗 Resources: https://aka.ms/iq-series
📌 This event is a part of a series, learn more here: https://aka.ms/microsoftiqlive
#MSFTReactor #learnconnectbuild #foundryiq #serverless #knowledgebases #agentdevelopment #retrieval #datagrounding #azureaifoundry #enterpriseknowledge
[eventID:27452] Composing Knowledge Bases That Reason Over Work, Business, and the Web](https://i.ytimg.com/vi/rvDTN3GGnm8/mqdefault.jpg)
![Microsoft Certifications Explained!
Microsoft certifications can feel like a maze of exams, role-based certifications, specializations, and constantly evolving technologies.
In this session, we’ll cut through the complexity and make sense of the Microsoft certification landscape.
Whether you’re starting your certification journey, looking to level up your career, or wondering which certification to pursue next, we’ll explore the different paths, what the certifications really mean, and how to choose the right ones for your goals.
We’ll also look at the impact of AI and the latest changes to Microsoft’s certification portfolio.
Come ready to rethink your certification strategy and leave with a clearer roadmap for your next Microsoft certification.
🔗 Learn more by exploring the resources: https://aka.ms/MicrosoftCredentialsOverview
0:00 Introduction and Speaker Presentation
3:00 Navigating Microsoft Learn & Certification Resources
5:25 Strategy: Choosing the Right Certification
8:53 Why Get Certified? (Purpose, Learning & Career)
13:05 Credentials Overview (Certifications vs. Applied Skills)
15:15 Fundamentals Certifications Overview
19:15 Specialty Certifications
20:00 Career Goal Planning & Roadmap Examples
25:05 AI Certification Focus (Complementary Skills)
27:00 The 8-Step Preparation Method
42:00 Exam Day Tips & Proctoring
50:00 Q&A and Specific Certification Paths
1:04:15 Conclusion
#MicrosoftReactor #LearnConnectBuild #Certifications #MSFTCertifications
[eventID:27598] Microsoft Certifications Explained!](https://i.ytimg.com/vi/sL-uMtgYhWY/mqdefault.jpg)
![Governing IQ-Powered Agents in Production with Agent 365
Learn how to operate and govern IQ-powered agents in production with Agent 365. In this session, well explore the challenges that arise as agents move beyond development and into real-world use, including visibility, policy enforcement, and lifecycle management.
Youll learn how Agent 365 works alongside the Microsoft IQ stack to help organizations monitor, manage, and govern agents at scale while maintaining trust, security, and operational control.
🔗 Resources: https://aka.ms/iq-series
📌 This event is a part of a series, learn more here: https://aka.ms/microsoftiqlive
#MSFTReactor #learnconnectbuild #foundryiq #serverless #knowledgebases #agentdevelopment #retrieval #datagrounding #azureaifoundry #enterpriseknowledge
[eventID:27457] Governing IQ-Powered Agents in Production with Agent 365](https://i.ytimg.com/vi/scnmeCNCk-0/mqdefault.jpg)
![AI で構築する: エージェント型ワークフローを構築するための GitHub Copilot SDK
セッション紹介
AI はコーディングを支援するだけではありません。複雑なタスクの自動化や、インテリジェントなワークフローのオーケストレーションにも活用できます。
このセッションでは、GitHub Copilot SDK を使用して、実際のエージェント駆動型ワークフローを構築する方法を学びます。
ハンズオン形式のデモを通じて、AI がツール、コンテキスト、複数ステップの実行を連携させながら、課題管理からコードの反復改善まで、日常的な開発タスクをどのように自動化できるかをご紹介します。
コードを書く段階を超え、インテリジェントでタスク主導のワークフローを設計する方法を学びましょう。
学習内容
- GitHub Copilot SDK を使用してエージェント型ワークフローを構築する方法
- エージェントが複数ステップのタスクを計画、推論、実行する方法
- エージェントをツール、サービス、外部システムと接続する方法
- 日常的な開発プロセスを自動化するインテリジェントなワークフローを設計する方法
使用するテクノロジー
- GitHub Copilot SDK
- エージェントランタイム(計画と複数ステップ実行)
- Model Context Protocol(MCP)
- ツール呼び出しとシステム統合
- ワークフローオーケストレーション
対象者
- エージェント駆動型ソリューションの構築に関心のある開発者
- GitHub Copilot SDK の機能を検討しているエンジニア
- インテリジェントなワークフローを構築するプラットフォームおよびツール開発者
- AI 支援コーディングからワークフロー自動化へステップアップしたい方
0:00:36 - セッション開始・自己紹介
0:01:45 - 本日の学習内容
0:02:53 - AI支援コーディングからエージェント型ワークフローへの変遷
0:06:12 - GitHub Copilot SDKの概要と主要要素
0:12:41 - 安全な実行のための隔離モデル
0:14:19 - デモアプリケーションの紹介(小売分析アシスタント)
0:17:38 - 小売分析アシスタントのデモ実演
0:24:34 - SDKにおけるストリーミングとイベント処理の実装解説
0:31:01 - ツールサンプルとイベントの流れの観察
0:35:08 - モデルの選択と動的な設定
0:41:23 - ツール定義とエージェントの判断プロセス
0:46:02 - 業務データとMCPサーバーの統合
0:53:13 - 全体アーキテクチャの振り返り
0:57:25 - セッションのまとめと今後のご案内
[eventID:27491] AI で構築する: エージェント型ワークフローを構築するための GitHub Copilot SDK](https://i.ytimg.com/vi/sd5AODNv9NU/mqdefault.jpg)