Uploaded July 2026 | Updated September 2026, 2 weeks ago
AI projects rarely fail because of the model — they fail because the data foundation underneath is fragile: agents that relearn the business every time, metrics that contradict each other, and no clear lineage for how an answer was reached.
This session breaks down Fabric IQ, Microsoft’s semantic and ontology layer for enterprise data, in plain terms. We’ll cover how Semantic Models, Ontologies, and Agents work together to turn raw tables into governed, trusted business context — enabling natural language interaction, consistent definitions across teams, and explainable AI answers. Attendees will leave with a clear, jargon-free understanding of where Fabric IQ fits in the modern data and AI stack, and how to start experimenting with it.
🔗 Learn more by exploring the resources: https://aka.ms/FabricIQ/overview
#MicrosoftReactor #LearnConnectBuild
[eventID:27482]
AI projects rarely fail because of the model — they fail because the data foundation underneath is fragile: agents that relearn the business every time, metrics that contradict each other, and no clear lineage for how an answer was reached.
This session breaks down Fabric IQ, Microsoft’s semantic and ontology layer for enterprise data, in plain terms. We’ll cover how Semantic Models, Ontologies, and Agents work together to turn raw tables into governed, trusted business context — enabling natural language interaction, consistent definitions across teams, and explainable AI answers. Attendees will leave with a clear, jargon-free understanding of where Fabric IQ fits in the modern data and AI stack, and how to start experimenting with it.
🔗 Learn more by exploring the resources: https://aka.ms/FabricIQ/overview
#MicrosoftReactor #LearnConnectBuild
[eventID:27482]
![Hands-On with Skills in Microsoft Copilot Studio
The session will introduce the concept of skills, explain how they differ from general agent instructions, and demonstrate how they enable agents to perform specialised tasks with greater accuracy and consistency.
Through a live demonstration, attendees will learn how to:
- Create and configure skills
- Define activation instructions
- Incorporate guidelines and examples
- Test skill behaviour in real-world scenarios
By the end of the session, participants will have a clear understanding of how skills can be leveraged to:
- Improve agent capabilities
- Promote reusable business logic
- Build intelligent AI solutions using Microsoft Copilot Studio
[eventID:27513] Hands-On with Skills in Microsoft Copilot Studio](https://i.ytimg.com/vi/9uzRcffBnSU/mqdefault.jpg)
![POSETTE: An Event for Postgres 2026 – Livestream 3
Mark your calendars for the Livestream 3 for POSETTE: An Event for Postgres 2026, happening Tue Jun 17, 2026. This livestream has 11 unique talks presented by Postgres experts & users. POSETTE is a free and virtual developer event with 44 talks delivered across 4 livestreams—organized by the Postgres team at Microsoft, in partnership with AMD. Detailed schedule with speakers and abstract for Livestream 3: https://posetteconf.com/2026/schedule/#livestream3
► Video chapters:
⏩ 00:00 – Waiting Room music & intro
⏩ 14:58 – Welcome with hosts Divya Bhargov & Abe Omorogbe
⏩ 27:29 – Livestream 3 Trailer
⏩ 32:57 – The Wonderful World of WAL, by Bruce Momjian
⏩ 1:02:29 – Building Event-Driven Systems with PostgreSQL Logical Replication and Drasi, by Diaa Radwan
⏩ 1:31:47 – pgcov: Bringing Real Test Coverage to PostgreSQL Code, by Pavlo Golub
⏩ 1:58:13 – Livestream 4 Trailer
⏩ 2:05:34 – Quorum-Based Consistency for Cluster Changes with CloudNativePG Operator, by Jeremy Schneider & Leonardo Cecchi
⏩ 2:36:57 – From Queries to Agents: The Next Era of Data Retrieval on PostgreSQL, by Abe Omorogbe
⏩ 3:04:00 – PostgreSQL 17 vs 18: Side‑by‑Side Performance Wins in Real‑World Queries, by Divya Bhargov
⏩ 3:31:08 – Production RAG at Scale with Azure Database for PostgreSQL, by Julia Schröder Langhaeuser & Paula Santamaria
⏩ 3:57:51 – Livestream 1 Trailer
⏩ 4:04:00 – The Rise of PostgreSQL as the Everything Database, by Varun Dhawan
⏩ 4:31:19 – Postgres isn’t slow, your storage is, by Sai Srirampur
⏩ 5:00:38 – Why we built Azure HorizonDB for PostgreSQL, by Dingding Lu
⏩ 5:22:05 – Maintaining Large Tables in PostgreSQL, by Sarat Balijepalli
⏩ 5:54:16 – Livestream 2 Trailer
⏩ 5:58:47 – Livestream 3 Wrap-Up with Divya Bhargov & Abe Omorogbe
There’s more to discover about POSETTE: An Event for Postgres 2026. Here are some helpful resources:
🌐 Everything you need to know about POSETTE: An Event for Postgres 2026: https://posetteconf.com/2026/
💬 The virtual hallway track during the Livestream will happen in the Microsoft Open Source Discord channel #posetteconf @ https://aka.ms/open-source-discord
📝 Take a look at the “Ultimate Guide to POSETTE (2026 edition)”
https://techcommunity.microsoft.com/blog/adforpostgresql/ultimate-guide-to-posette-an-event-for-postgres-2026-edition/4520246
You can also follow @PosetteConf on social media for even more updates:
LinkedIn: https://www.linkedin.com/company/posetteconf/
X: https://x.com/PosetteConf
Mastodon: https://mastodon.social/@posetteconf
Bluesky: https://bsky.app/profile/posetteconf.com
Search for #PosetteConf
Come learn what you can do with PostgreSQL, the world’s most advanced open-source relational database—from the nerdy to the sublime.
#PosetteConf #PostgreSQL #database
[eventID:26855] POSETTE: An Event for Postgres 2026 – Livestream 3](https://i.ytimg.com/vi/9yzpwi50uLA/mqdefault.jpg)
![Build GitHub Skills: Custom Self-Paced Learning with the gh-skills-builder Plugin
Great developer training shouldnt require a team of curriculum engineers.
In this hands-on Microsoft Reactor series, youll learn to design, build, review, and publish your own **GitHub Skills-style exercises** — interactive, self-paced, issue-driven learning experiences that live right inside a GitHub repository — using the open-source **gh-skills-builder** GitHub Copilot plugin. Across the series well take a raw idea, workshop, or demo and turn it into a complete, validated learning journey.
Youll see how the plugins **custom agents** and **Agent Skills** collaborate to handle each stage of the exercise lifecycle, so you can focus on *what* you want learners to master instead of the plumbing behind it.
🔗 Learn more by exploring the resources: https://aka.ms/gh-skills-builder
#MicrosoftReactor #LearnConnectBuild #AgentSkills #GitHubCopilot
[eventID:27558] Build GitHub Skills: Custom Self-Paced Learning with the gh-skills-builder Plugin](https://i.ytimg.com/vi/AIPfnSEoZyg/mqdefault.jpg)

![Secure, Expose, and Manage Your APIs with Azure API Management
Modern applications are API-driven and managing those APIs is critical for scale, security, and governance.
In this episode, you’ll learn how to: - Publish and secure APIs across your applications - Apply policies for authentication, throttling, and governance - Manage internal and external API consumption - Create a unified API layer across ACA, AKS, and App Service
🔗 https://aka.ms/BCNA/API/appservice
📌 This event is a part of a series, learn more here: https://aka.ms/BCNA
[eventID:27397] Secure, Expose, and Manage Your APIs with Azure API Management](https://i.ytimg.com/vi/AaVHj9en30g/mqdefault.jpg)
![Fabric Data Agent como servidor MCP: expón tus datos empresariales a cualquier LLM
Esta sesión es el núcleo de la serie: aprenderás a convertir un Fabric Data Agent en un servidor MCP para que cualquier herramienta de IA externa — Claude, Copilot, agentes custom — pueda consultarlo sin integraciones personalizadas.
Crearemos un Data Agent sobre datos reales en Fabric, lo publicaremos como servidor MCP generando su URL y archivo mcp.json, y lo conectaremos desde VS Code.
Veremos cómo el agente recibe preguntas en lenguaje natural, las traduce a consultas sobre el lakehouse o warehouse y devuelve respuestas gobernadas. Cubriremos requisitos de capacidad (F2+), límites de compliance y configuración para respuestas consistentes.
Ten preparado un Fabric Data Agent básico con al menos una tabla conectada — en la sesión anterior damos las instrucciones.
Consumir el agente de datos de Fabric como un servidor del protocolo de contexto del modelo
https://learn.microsoft.com/fabric/data-science/data-agent-mcp-server/?wt.mc_id=youtube_27353_organicsocial_reactor
📌 Este evento es parte de una serie, aprende más aquí: https://aka.ms/AgentesFabric
0:00 Introducción y presentación del tema
4:37 ¿Qué es el *Fabric Data Agent*?
7:25 ¿Cómo funciona el *Fabric Data Agent*?
10:42 Arquitectura técnica del *Data Agent*
16:52 Importancia del modelo semántico y preparación de datos
22:25 Mejores prácticas para la implementación
28:28 Configuración de la capacidad paga (Requisitos)
31:51 Creación del *Data Agent* en el espacio de trabajo
35:46 Creación del *Lakehouse* y Notebooks
42:36 Documentación y enriquecimiento del modelo semántico
49:00 Preparación de datos para la IA (Copilot)
53:20 Configuración de respuestas verificadas
1:01:41 Conexión y publicación del agente (MCP)
1:06:57 Demo: Interacción con el agente mediante lenguaje natural
[eventID:27353] Fabric Data Agent como servidor MCP: expón tus datos empresariales a cualquier LLM](https://i.ytimg.com/vi/AeLxhYtQTX4/mqdefault.jpg)

![Automations, Agents & Backlogs: Inside the GitHub Copilot App
In this episode of New Breakpoint, Michelle Sandford sits down with Aaron Powell to explore the GitHub Copilot App - a new experience that brings agentic workflows directly into your day-to-day work on GitHub. Rather than focusing purely on code, the Copilot App introduces a shift toward task-driven collaboration, enabling developers and broader technical teams to manage pull requests, issues, and backlogs through intelligent, agent-powered workflows.
Together, Michelle and Aaron take a hands-on look at how the app enables:
- Automations: creating scheduled workflows like daily pull request summaries or issue triage, tailored to your team’s needs
- Agent-driven sessions: working across repositories with isolated contexts, allowing multiple streams of work to run in parallel
- Backlog visibility: helping product owners and managers stay across issues and progress without needing to dive into code
- Interactive collaboration: reviewing changes, adding inline feedback, and steering agents in real time
- Extensibility: using skills, plugins, and canvases to customise workflows, visualise data, and even build lightweight apps
The episode also explores how the Copilot App bridges the gap between developers and non-developers, making it easier for entire teams to participate in modern, AI-powered workflows - whether you’re writing code, managing a sprint, or tracking progress on the go. If you’re looking to understand how automations, agents, and real-world workflows come together inside GitHub, this is a great place to start.
Get started with the GitHub Copilot app: https://awesome-copilot.github.com/learning-hub/github-copilot-app/
Learn more about the series: https://aka.ms/NewBreakpoint/r
0:04 Welcome to New Breakpoint
0:28 Meet Aaron Powell and GitHub Copilot
2:07 Copilot SDK vs GitHub Copilot App
3:22 What Is the GitHub Copilot App?
5:19 First Look at the Copilot App Interface
7:13 Automations and Daily GitHub Summaries
9:30 Building Custom Automations
11:32 Copilot for Product Owners and Technical Teams
13:43 Sessions, Agents, and Workflow Management
17:06 Git Work Trees and Parallel Agent Workflows
19:30 Merge Conflicts and Multi-Agent Collaboration
21:18 Creating Sessions from Pull Requests
24:02 Code Reviews and Context-Aware Comments
28:10 Session Navigation and History Tracking
29:29 Opening Copilot Sessions in VS Code
32:27 Skills and Extensions in Copilot App
32:58 Introducing Copilot Canvas
34:21 Building Custom Canvas Experiences
35:28 Canvas Demo: Building a Snake Game
36:38 Integrating External Services and Azure DevOps
38:48 Awesome Copilot and the Learning Hub
41:05 Custom Notifications and Sound Effects
42:26 Documentation and Best Practices
44:58 How to Get the GitHub Copilot App
45:36 Mobile Access and Cloud Agents
47:05 Managing Agent Sessions from GitHub Mobile
47:42 Real-World Cloud Agent Workflow
49:41 Final Thoughts on Developer Productivity
50:07 Resources and Further Learning
50:46 Live Q&A Session
[eventID:27339] Automations, Agents & Backlogs: Inside the GitHub Copilot App](https://i.ytimg.com/vi/BLSTfoOsajs/mqdefault.jpg)


![OpenClawNet - Automatización + Azure + Foundry
En esta sesión prepararemos el agente para producción. Agregaremos programación de tareas como servicio de fondo, integraremos Azure OpenAI y Azure AI Foundry como proveedores de modelos en la nube, construiremos dashboards de monitoreo, y escribiremos tests unitarios que verifiquen todo el stack.
Al final tendremos una plataforma de agentes completa, testeada y lista para la nube con manejo de errores de nivel producción, múltiples proveedores de modelos AI, automatización programada, y dashboards de observabilidad. Despliégala en Azure Container Apps, App Service, o ejecútala on-premises con Ollama.
Tópicos y Tecnologías Clave:
- BackgroundService para tareas de larga duración
- Integración de Azure OpenAI y selección de modelos
- Azure AI Foundry (inferencia, despliegues, evaluación)
- Patrones de testing unitarios para agentes y herramientas
- Health checks y diagnósticos del sistema
- Job queuing y tracking de estado
- Configuración para despliegues multi-ambiente (local, staging, producción)
📌 Este evento es parte de una serie, aprende más aquí: https://aka.ms/NET10GHC/es/y
Capítulos:
0:00 Introducción y presentación de OpenClaw.net
1:27 ¿Qué es OpenClaw.net? Propósito del proyecto
2:47 Agenda: Aspire, Trazas, Skills y Secretos
4:00 Recorrido por el repositorio público
4:45 Flujo de trabajo con GitHub Copilot (PRs y Code Review)
7:41 Gestión de múltiples repositorios y sincronización (Privado vs Público)
11:42 Análisis de Pull Requests con Copilot
16:03 Uso de Canvases en GitHub Copilot
21:19 Aspire en Visual Studio Code: Dashboard y monitorización
24:43 Foundry Tool Kit para la gestión en la nube
29:07 Implementación de Secret Vault (Bóveda de secretos)
32:38 Depuración en vivo con Copilot y el navegador integrado
36:46 Demostración de pruebas con Playwright
41:30 Configuración de agentes de sistema y perfiles
42:37 Gestión de memoria y retrieval en agentes
45:04 Funcionamiento de las Skills y carga de archivos
48:17 Prueba práctica: Generación de contenido con Skills y búsqueda online
51:20 Ciclo de vida y borrado de Skills
53:18 Implementación de comandos en chat: El desafío de stop
56:30 Conclusiones y próximos pasos (Despliegue)
#microsoftreactor #learnconnectbuild
[eventID:26926] OpenClawNet - Automatización + Azure + Foundry](https://i.ytimg.com/vi/BY94TELQv7o/mqdefault.jpg)