Microsoft Developer
Microsoft DeveloperVideo
- Format:
- YouTube channel
- New videos:
- Several times a week
- Typical length:
- about 10 minutes, not counting Shorts
- Latest video:
About this show
Official Microsoft Developer YouTube channel! Featuring live events, tips and tutorials, product news and more to help you explore, discover, and build solutions - using your favorite languages, technologies, and open-source frameworks.
More information visit https://developer.microsoft.com
Covers: software development, AI agents, microsoft azure, cloud computing, microsoft foundry, database performance, developer tools, azure sql database
Videos come from the channel's public YouTube feed. The artwork is the newest video thumbnail. Last updated . Names, artwork and text belong to their publishers.
Recently on this show
Summary of recent videos, drawn from the publisher's feed.
Microsoft Developer's channel covers tools, tutorials and product news for building software with popular languages, technologies and open-source frameworks, plus live events. Recent videos include Azure HorizonDB in VS Code, choosing smaller AI models over bigger ones, testing AI agents before shipping, and fuzzy matching for search.
Cosmos
Similar in subject
- Amazon Web Services
Both cover: cloud computing, developer tools, software development - Google Cloud Tech
Both cover: cloud computing, developer tools, software development - Microsoft Azure
Both cover: cloud computing, software development, ai agents
Similar shows
Videos
Azure HorizonDB + VS Code: Your AI App Development Workspace
· 1 min
Build and run Azure HorizonDB right from Visual Studio Code. Go from setup to intelligent apps without leaving your workflow. Azure HorizonDB is one platform for transactions, vectors, AI and orchestration, helping your team build smarter apps faster Learn more: https://azure.microsoft.com/en-us/products/horizondb
Not rated yetMicrosoft Foundry: Bigger Models Aren’t Always Better
· 39 sec
Microsoft Foundry puts a full catalog of models in one place, so match the model to the job. Adam Bogobowicz and Adrian Twarog on why a smaller model can do the same task for less. From our full conversation: https://www.youtube.com/watch?v=-B9pIyQfaSU
Not rated yetMicrosoft Foundry: Keep Your AI Spend Under Control
· 45 sec
Microsoft Foundry runs inside Azure, so you can see every model, credit and dollar of spend in one place. Adrian Twarog on why that matters before agents run wild. From our full conversation: https://www.youtube.com/watch?v=-B9pIyQfaSU
Not rated yetBuild smarter search without AI: Fuzzy matching for applications | Data Exposed
· 18 min
Not every search problem requires AI, embeddings, or vector databases. Most business applications simply need to handle real-world data imperfections: misspelled customer names, inconsistent product descriptions, duplicate records, incomplete entries, and user input that rarely matches stored data exactly. Traditional string-matching techniques such as LIKE, CHARINDEX, PATINDEX, Full-Text Search, and even Regular Expressions solve many search scenarios, but they often struggle when the data itself is imperfect. In this session, we'll explore the newest fuzzy string-matching capabilities in Azure SQL and SQL Server and see where they fit within the broader SQL search ecosystem. Through practical demonstrations, we'll compare fuzzy matching against existing approaches including LIKE predicates, Full-Text Search, and Regular Expressions, helping you understand the strengths, limitations, and ideal use cases for each technique. Using real-world scenarios such as customer lookup, product search, duplicate detection, master data management, data cleansing, and application search experiences, you'll learn how to build typo-tolerant and user-friendly applications without introducing unnecessary AI complexity. ✅ Let's connect: LinkedIn - Anna Hoffman, https://www.linkedin.com/in/amthomas46/ Twitter - AzureSQL, https://aka.ms/azuresqltw LinkedIn - Abhiman Tiwari, https://www.linkedin.com/in/abhiman-tiwari/ 🔴 Watch even more Data Exposed episodes: https://aka.ms/dataexposedyt 🔔 Subscribe to our channels for even more SQL tips: Microsoft SQL / Azure SQL: https://aka.ms/msazuresqlyt Microsoft SQL Server: https://aka.ms/mssqlserveryt Microsoft Developer: https://aka.ms/microsoftdeveloperyt #AzureSQL #SQL #LearnSQL
Not rated yetYour Path to Production for Agents
· 1 hr 2 min
Building an AI agent that impresses as a demo has never been easier. Building one that is secure, governable, cost-effective, and capable of delivering consistent business value in production is a very different challenge. In this first episode in our Path to Production for Agents master class series, Adelina Balasa, AI Global Strategy Lead at Microsoft, shares a practical framework developed from working with hundreds of customers and partners, helping organizations navigate the journey from AI experimentation to enterprise-scale deployment. Rather than focusing solely on models, the series explores the architectural, operational, security, and organizational foundations required to successfully move agents from concept to production and beyond. The Path to Production for Agents series is built around six essential pillars that help organizations overcome the most common barriers to production success: AI Center of Excellence, AI Governance, AI Landing Zones, AI Security & Trust, Agent Ops, and Solution Optimization. Together, these pillars provide a roadmap for establishing the right people, processes, platforms, and practices to deploy and scale AI responsibly and efficiently. Each episode dives deeper into one of these pillars, offering practical guidance and actionable steps that teams can apply immediately. What You'll Learn in this Episode - Why so many AI projects succeed as demos but struggle to make the leap into production. - The common challenges organizations encounter when moving agents from experimentation to deployment and scale. - Why production readiness is typically an architectural and organizational challenge, not simply a model selection problem. - An introduction to the six pillars of the Adelina Balasa Path to Production framework and how they work together to accelerate AI transformation. - How an AI Center of Excellence helps align people, processes, and best practices for successful AI adoption. - The role of AI Governance, AI Landing Zones, and AI Security & Trust in creating a secure and scalable foundation for AI. - How Agent Ops and Solution Optimization help improve deployment speed, operational reliability, performance, and cost efficiency. - How to use the Path to Production framework as a practical roadmap for taking AI agents from ideation to real-world business impact. Connect - Adelina Balasa https://www.linkedin.com/in/adelina-balasa-b2006a67/ Related episodes - Watch the entire Path to Production series https://aka.ms/PathToProductionForAgents Chapters 00:00 Introduction 00:56 Demo to Production 02:50 AI Journey 03:11 Production Gap 03:37 Framework Overview 04:29 AI CoE 13:39 Governance 23:16 AI Landing Zones 32:00 AU Security and Trust 38:39 AgentOps 45:42 Solution Optimization 53:33 Pulling it all together 54:37 Agent Architectures 1:00:00 Remember these three things
Not rated yetMicrosoft Foundry: Test Your Agent Before You Ship
· 34 sec
Microsoft Foundry’s playground is where you test an agent before you publish it. Adrian Twarog builds one and checks it with a single prompt. From our full conversation: https://www.youtube.com/watch?v=-B9pIyQfaSU
Not rated yetWeb IQ: Grounding Agents with Real World Intelligence
· 22 min
Discover how Microsoft Web IQ helps AI agents access fresh, trustworthy information from across the web, and how grounding, passage-level retrieval, citations, and scopes improve answer quality and transparency. In this session we'll explore Web IQ’s search verticals, Auto mode, Places, and fast versus advanced retrieval. See how its low-latency, token-efficient architecture supports production-scale agents. Chapters: 0:00 - Why Agents Need Web Grounding 1:36 - Introducing Microsoft Web IQ 2:12 - Grounded Answers with Citations 3:17 - Risks of Outdated Information 4:22 - What Is Web IQ? 5:30 - Search Modes and API Options 6:36 - Quality, Speed, and Token Efficiency 8:15 - Passage-Level Retrieval 9:22 - Reduce Agent Latency and Costs 10:26 - Responsible Enterprise Grounding 11:29 - Control Sources with Scopes 12:34 - Auto Search and Places 14:47 - Fast vs. Advanced Mode 15:52 - Competitive Intelligence Example 17:31 - Set Up Your First Web IQ Call 18:37 - Build a Web-Grounded Agent 19:11 - Demo: Analyze US Energy Markets 20:18 - Inspect Activity and Citations 21:22 - Key Benefits and Resources Resources: Check out the Web IQ Cookbook in the repo: https://aka.ms/iq-series https://www.microsoft.com/en-us/webiq https://learn.microsoft.com/en-us/web-iq/
Not rated yetMicrosoft Foundry with Adrian Twarog: Build Your First Agent
· 15 min
Microsoft Foundry is where you build, test and publish AI agents, with usage and billing tracked in Azure. Developer and YouTube creator Adrian Twarog (@AdrianTwarog) takes Microsoft Foundry from first login to a published agent. He creates an agent, tests it in the playground, deploys models, connects an MCP server with scoped permissions and calls the finished agent from the Azure SDK. Along the way, he shows the model catalog, where models from Microsoft, OpenAI, Anthropic and DeepSeek sit side by side. Then he sits down with Adam Bogobowicz from Microsoft to talk about what it really took to make it look easy. In the conversation, Adrian explains why he goes through a product three times before he records, why creating an agent in Microsoft Foundry also deploys a model, and why seeing your models, credits and spend in one place matters for enterprise teams. This episode is for developers and IT teams trying Microsoft Foundry for the first time, and for anyone who needs to keep AI costs visible as agents move toward production. Resources - Microsoft Foundry portal https://ai.azure.com - More Azure resources! https://azure.com/AzureEssentials Connect - Adrian Twarog: https://www.youtube.com/c/AdrianTwarog - Adam Bogobowicz: https://www.linkedin.com/in/adambogobowicz Chapters 00:00 Why Adrian Twarog took on Microsoft Foundry 02:00 Set up a Microsoft Foundry account 03:53 Build, test and deploy your first agent 07:01 Connect an MCP server with scoped permissions 09:55 Publish your agent and call it from code 10:40 How Adrian makes a complex tool look simple
Not rated yetCopilot Harness, DirectX, and Azure Auth
· 3 min
This week on The Upload: Microsoft’s Copilot harness enters public preview in Foundry Agent Service, giving developers a managed agent loop and sandbox. The October Azure SDK release brings auth improvements across .NET, Python, and Java, plus a stable Cosmos DB client for Rust. And PIX adds console-style crash postmortems for Windows games. 📅 .NET Conf is November 10-13. We’ll be live from this studio, so tune in! #Microsoft #Azure #GitHubCopilot #DotNetConf #DevNews
Not rated yet52% More Throughput. 1.5x Better Value. The Azure SQL DB Hyperscale Advantage | Data Exposed
· 5 min
Learn about the new Azure SQL Database Hyperscale benchmark against Amazon Aurora. "To help organizations with their cloud selection process, the Principled Technologies team put Azure SQL Database Hyperscale and Amazon Aurora PostgreSQL I/O-Optimized instances to the test with an I/O-intensive HammerDB workload. Across two vCore counts and multiple concurrency loads, Azure SQL Database Hyperscale offered similar or stronger database performance than Amazon Aurora PostgreSQL—most notably at 16 vCores, where it delivered up to 52 percent more performance and 51 percent better performance per dollar." From the benchmark study: https://aka.ms/hyperscalebenchmark ✅ Let's connect: LinkedIn - Ravinder Singh, https://www.linkedin.com/in/ravinderpsingh/ LinkedIn - Anna Hoffman, https://www.linkedin.com/in/amthomas46/ Twitter - AzureSQL, https://aka.ms/azuresqltw 🔴 Watch even more Data Exposed episodes: https://aka.ms/dataexposedyt 🔔 Subscribe to our channels for even more SQL tips: Microsoft SQL / Azure SQL: https://aka.ms/msazuresqlyt Microsoft SQL Server: https://aka.ms/mssqlserveryt Microsoft Developer: https://aka.ms/microsoftdeveloperyt #AzureSQL #SQL #LearnSQL
Not rated yetAI That Works Offline
· 31 sec
An OT virtual coworker runs on site, so it keeps working even when the cloud connection drops. Jason Farmer from Microsoft's digital operations team explains. From our full conversation on physical AI architecture: https://www.youtube.com/watch?v=J9GDIAewjPY #PhysicalAI #EdgeAI #AzureEssentialsShow
Not rated yetWhat does it really take to ship an AI agent?
· 50 sec
Full episode: https://youtu.be/gF2RMvXput8 Foundry portal: https://aka.ms/foundry-portal Meet Sparkles, a cupcake-ordering agent built with Microsoft Foundry and Microsoft Agent Framework. Follow the full development cycle: connect to a deployed model, load the shop’s persona and welcome banner from MCP prompts, call live Cupcake Store tools, inspect execution traces, and evaluate the resulting conversations.
Not rated yetHow to Run AI Agents at the Industrial Edge
· 3 min
Physical AI connects real equipment to AI agents through a hybrid architecture. Cloud-connected devices link directly to Azure IoT Hub, while on-site devices run through Azure IoT Operations on Kubernetes at the edge, with Foundry Local running AI workloads on premises. Data from both paths lands in Microsoft Fabric, and Azure Resource Manager manages it all. Jason Farmer, Product Manager on Microsoft's digital operations team, walks through the architecture and the reason it exists: virtual coworkers. An OT virtual coworker runs on site and keeps working even when the cloud connection drops, while an enterprise virtual coworker in Microsoft Foundry or Fabric reaches frontline and information workers through Teams and Dynamics. Resources - Demo: Making industrial AI practical for real-world operations with adaptive cloud https://aka.ms/demo-industrialai - Explore more Essential resources! https://azure.com/AzureEssentials Connect - Jason Farmer https://www.linkedin.com/in/jason-farmer-9492b44/ Chapters 0:00 What physical AI needs at the edge 0:40 How cloud and on-site devices connect to Azure 1:25 How physical AI data flows into Microsoft Fabric 1:55 What virtual coworkers do for frontline teams 2:20 How an OT virtual coworker keeps working offline 3:05 Where to see physical AI in action
Not rated yetWebTransport Headers, Zstd WAL Compression, and Copilot Studio Conversation Harness
· 2 min
A browser change that can alter runtime behavior without touching your code, a storage-saving option for Azure Database for PostgreSQL, and a new way to build customer-facing agents in Copilot Studio. Missed GitHub Universe? Catch up on the keynotes and sessions by subscribing to the GitHub YouTube channel (https://www.youtube.com/@GitHub) and the Microsoft Developer YouTube channel (https://www.youtube.com/@MicrosoftDeveloper).
Not rated yetBuild an AI agent without leaving VS Code!
· 1 min
Full episode: https://youtu.be/lLTqWYDhHZw Foundry portal: https://aka.ms/foundry-portal Build the Sparkles agent using the Microsoft Foundry Toolkit. Configure the deployed model and MCP connection, invoke the agent from the development environment, and verify that it can use the shop’s prompts and live ordering tools. Store traces, evaluation datasets, and results with the project for repeatable testing.
Not rated yetRunning a 120B model offline on Surface RTX Spark Dev Box with Intelligent Terminal
· 8 min
See what's possible with Intelligent Terminal running a local model on Surface Laptop Ultra. A 120b parameter local models runs entirely locally on the laptop's NVIDIA RTX Spark N1X with 128 GB of unified memory. See how these powerful new devices can complete complex flows such as writing new agent skills with context all without going to the cloud. Learn more about Surface Laptop Ultra: Surface Laptop Ultra: https://www.microsoft.com/en-us/surface/devices/surface-laptop-ultra and Surface RTX Spark Dev Box: Surface RTX Spark Dev Box: https://www.microsoft.com/en-us/surface/devices/surface-rtx-spark-dev-box Get Intelligent Terminal on GitHub: https://github.com/microsoft/intelligent-terminal or the Microsoft Store: https://apps.microsoft.com/detail/9nmqc2ssjx24
Not rated yetAutomation vs Physical AI
· 37 sec
Automation repeats a task; physical AI gets smarter every time it runs. Michael MacKenzie, VP of Product Management for Azure IoT and Digital Operations at Microsoft, explains. From our full conversation on IoT and physical AI: https://www.youtube.com/watch?v=NrcM8Rq73Q4 #PhysicalAI #IndustrialAI #AzureEssentialsShow
Not rated yetCustomize your agent from a Chat Window?
· 1 min
Full episode: https://youtu.be/ojQu4RwvvIA Visit: https://aka.ms/foundry-portal What if the biggest model is quietly the wrong one? Using the Microsoft Foundry model catalog, we understand the best model for our use case. We swap only the model deployment. We run the same evaluation over both. We also capture cost and latency from real runs. That turns “pick a model by vibes” into a measured decision. By the end, you’ll see when a smaller model can be the production choice. You’ll also see when it isn’t.
Not rated yetStop wiring infrastructure, start building agents
· 1 min
Full episode: https://youtu.be/AMxLLFeUyRM Foundry portal: https://aka.ms/foundry-portal Introducing the Microsoft Copilot harness! Every agent project seems to start the same way containers, scaling, state, auth. Most of the time, you don't need any of it. You rebuild Sparkles as a prompt agent: a model, instructions, and the Cupcake Store MCP tools, defined once and versioned in Microsoft Foundry. No container, no server, no pipeline. Then you hit the edge where a prompt agent isn't enough, such as custom orchestration or your own framework, and move the same agent to a hosted agent. Callers still invoke it by name. By the end, you'll know which one to reach for first, and when to switch. #copilot #harness #foundry #microsoft #ai
Not rated yetStop wiring infrastructure, start building agents
· 6 min
https://aka.ms/foundry-portal https://aka.ms/InsideMicrosoftFoundryPlaylist Every agent project seems to start the same way containers, scaling, state, auth. Most of the time, you don't need any of it. You rebuild Sparkles as a prompt agent: a model, instructions, and the Cupcake Store MCP tools, defined once and versioned in Microsoft Foundry. No container, no server, no pipeline. Then you hit the edge where a prompt agent isn't enough, such as custom orchestration or your own framework, and move the same agent to a hosted agent. Callers still invoke it by name. By the end, you'll know which one to reach for first, and when to switch. 0:00 - Scale Complex Cupcake Orders 0:34 - Introducing the Copilot Harness 1:07 - Configure Harness Tools 1:38 - Add Reusable Agent Skills 2:44 - Enable Bulk and Party Orders 3:18 - Process Orders from CSV Data 4:25 - Configure the Agent with YAML 4:59 - Define Skills and MCP Tools 5:32 - Scale Without Managing Infrastructure #copilot #harness #microsoft #foundry
Not rated yet