Uploaded June 2026 | Updated September 2026, 3 weeks ago
AI is changing how we interact with data, and you don’t need a complex setup to start building something powerful.
In this session, we’ll kick off a new SQL + AI community contest and show you exactly what it takes to participate and succeed. You’ll learn how modern AI patterns—from prompting and natural language interactions to embeddings, vector search, and retrieval-augmented generation (RAG)—connect back to SQL and real data.
We’ll focus on the practical side of building: how to take an idea, shape it with the right prompts, and turn it into something useful, whether that’s a simple interaction or a fully working intelligent app. Along the way, we’ll explore example scenarios, common patterns, and ways to think about solving real problems with SQL + AI.
This session is designed to get you ready. By the end, you’ll understand what the contest expects, how to approach it, and how to apply these concepts beyond the contest—in your own projects and real-world work!
📌 This event is a part of a series, learn more here: https://aka.ms/DataDays/cert-y
Register for Data Days: https://aka.ms/datadays
Chapters:
0:00 Introduction and Welcome
0:30 About the Speakers
2:28 Community Events and Certifications
8:47 Behind the Scenes: Moderators
9:59 Why Data Agents?
11:21 Grounding AI in Enterprise Knowledge
13:30 Fabric IQ and Ontologies
15:08 Fabric Data Agents in M365
17:15 Demo: M365 Copilot Integration
19:50 Integration with Copilot Studio
20:44 Azure AI Foundry
22:58 General Availability of Data Agents
24:52 Demo: Zava Trade Expert and Graph Data
30:26 Graph DB Support
31:07 Audit Logs and Governance
36:13 Security and Least Privilege
37:45 Code Interpreter
39:21 Visuals and Data Agents
42:13 Assisted Setup Mode
43:38 Optimized SQL Query Generation
44:40 Architectural Design
49:52 MCP Server and Client
54:01 Data Science: AI Functions and AutoML
55:27 ML Model Monitoring
56:44 MLflow and Model Governance
57:44 Data Science Roadmap
59:00 Data Agents Roadmap
1:00:00 Resources and Getting Started
1:01:21 Conclusion and Global Fabric Day
#microsoftreactor #learnconnectbuild #DataDays
[eventID:27330]
AI is changing how we interact with data, and you don’t need a complex setup to start building something powerful.
In this session, we’ll kick off a new SQL + AI community contest and show you exactly what it takes to participate and succeed. You’ll learn how modern AI patterns—from prompting and natural language interactions to embeddings, vector search, and retrieval-augmented generation (RAG)—connect back to SQL and real data.
We’ll focus on the practical side of building: how to take an idea, shape it with the right prompts, and turn it into something useful, whether that’s a simple interaction or a fully working intelligent app. Along the way, we’ll explore example scenarios, common patterns, and ways to think about solving real problems with SQL + AI.
This session is designed to get you ready. By the end, you’ll understand what the contest expects, how to approach it, and how to apply these concepts beyond the contest—in your own projects and real-world work!
📌 This event is a part of a series, learn more here: https://aka.ms/DataDays/cert-y
Register for Data Days: https://aka.ms/datadays
Chapters:
0:00 Introduction and Welcome
0:30 About the Speakers
2:28 Community Events and Certifications
8:47 Behind the Scenes: Moderators
9:59 Why Data Agents?
11:21 Grounding AI in Enterprise Knowledge
13:30 Fabric IQ and Ontologies
15:08 Fabric Data Agents in M365
17:15 Demo: M365 Copilot Integration
19:50 Integration with Copilot Studio
20:44 Azure AI Foundry
22:58 General Availability of Data Agents
24:52 Demo: Zava Trade Expert and Graph Data
30:26 Graph DB Support
31:07 Audit Logs and Governance
36:13 Security and Least Privilege
37:45 Code Interpreter
39:21 Visuals and Data Agents
42:13 Assisted Setup Mode
43:38 Optimized SQL Query Generation
44:40 Architectural Design
49:52 MCP Server and Client
54:01 Data Science: AI Functions and AutoML
55:27 ML Model Monitoring
56:44 MLflow and Model Governance
57:44 Data Science Roadmap
59:00 Data Agents Roadmap
1:00:00 Resources and Getting Started
1:01:21 Conclusion and Global Fabric Day
#microsoftreactor #learnconnectbuild #DataDays
[eventID:27330]
![Build Your First Microsoft Foundry Agent: Executive Briefings from Technical Updates
Build and deploy a focused AI agent that turns complex technical updates into concise, business-ready executive briefings. Starting with a hosted-agent scaffold in VS Code, you will define agent instructions, test realistic incident scenarios in Agent Inspector, and deploy the completed agent to Microsoft Foundry Agent Service. Learnings: How to scaffold a Python hosted agent with the Foundry extension How to test agent behavior locally, including safety-boundary prompts How to deploy and validate the agent in the VS Code and Foundry playgrounds.
📌 This event is part of a series. Learn more here: https://aka.ms/FoundryToolkitVSCode/series
🔗 Learn more by exploring the resources:
https://aka.ms/FoundryToolkit/VSCode
https://aka.ms/FoundryToolkit/VSCode/Overview
0:06 - Introduction and Housekeeping
0:56 - Speaker Introduction
2:49 - Introduction to Microsoft Foundry Toolkit for VS Code
3:39 - Four Main Steps to Building a Hosted Agent
8:17 - Overview of the Executive Summary Agent Project
12:40 - Setting up the Foundry Toolkit in VS Code
14:20 - Creating an Agent from Templates
21:44 - Customizing Agent Instructions and Adding Tools
24:35 - Installing Dependencies and Environment Setup
27:26 - Local Debugging and Testing with Agent Inspector
31:45 - Deploying the Agent to Microsoft Foundry
34:47 - Verifying and Testing the Deployed Agent
36:06 - Publishing the Agent
38:52 - Exploring Foundry Local Model Options
41:15 - Comparison: Foundry vs. Copilot Studio
45:43 - Conclusion and Upcoming Sessions
[eventID:27523] Build Your First Microsoft Foundry Agent: Executive Briefings from Technical Updates](https://i.ytimg.com/vi/e4bhjB8YNGU/mqdefault.jpg)
![Context Is Everything: Building Agents That Know Your Work Context
Learn how Work IQ helps agents understand the people, projects, and information that matter most in your organization.
See how richer context can make your agents more relevant, grounded, and effective in everyday work.
🔗 Copilot Developer Camp: https://aka.ms/copilotdevcamp
📍 Learn more about the series:https://aka.ms/MS365/CopilotPlaybook/r
00:00 Welcome & Series Introduction
01:49 Understanding Work IQ and Microsoft 365 Context Intelligence
04:07 Choosing Between A2A, MCP, and REST Integration Patterns
06:06 Work IQ CLI and Developer Tools Overview
08:34 Live Demo: Building a Cat Cafe Agent with Work IQ
13:36 From Teams Chat to Agent Specification and Code Generation
22:21 Exploring the Generated Agent, Skills, and MCP Integration
25:42 Provisioning, Testing, and Using the Agent in Copilot
29:19 Copilot Co-Work Automation, Skills, and Weekly Briefings
35:15 Evaluations, Publishing, Monitoring, and Agent Lifecycle Management
42:44 Q&A: MCP, A2A, Logic Apps, Agent Toolkit, and Work IQ Servers
#MSFTReactor #learnconnectbuild #developer #workiq #agents #copilot
[eventID:27450] Context Is Everything: Building Agents That Know Your Work Context](https://i.ytimg.com/vi/eEXW-sUAnSs/mqdefault.jpg)
![Modern Data Engineering for AI Applications
AI applications depend on scalable and reliable data platforms.
This session explores how Microsoft Fabric enables modern data engineering workflows through lakehouse architecture, data ingestion, orchestration, Spark-based processing, analytics, and governance.
Attendees will learn how to build strong data foundations to support analytics, machine learning, and generative AI workloads.
Key takeaways:
- Modern lakehouse architecture
- Building scalable data pipelines
- Spark and Fabric integration
- Data engineering best practices for AI workloads
00:00 Welcome & Housekeeping
01:41 Speaker Introduction & Background
03:23 AI Language Bias and the English-First Problem
06:19 Why Multilingual AI Matters
08:59 Case Study: Document Extraction Across Languages
11:58 Challenges in Non-English NLP
14:37 Architecting Multilingual AI Solutions
19:07 Common Pitfalls: Language Identification & Data Conversion
25:07 Processing Non-Latin Characters
27:40 Evaluating Multilingual AI Systems
29:54 Key Takeaways & Transition to Q&A
32:15 Introduction to FLAIR and DRI Copilot
36:04 Why Retrieval Systems Fail Over Time
38:20 Historical Analysis of User Queries
40:50 FLAIR: Using Feedback to Improve Retrieval
43:21 Offline Preprocessing & Signal Generation
46:28 Online Retrieval and Reranking
50:31 Computing Vote Scores from User Feedback
56:04 Evaluation Methodology & Results
1:01:45 Audience Q&A
1:03:50 Closing Remarks & Survey
📌 This is part of a series, learn more here: https://aka.ms/ProdReadySystems/series
[eventID:27360] Modern Data Engineering for AI Applications](https://i.ytimg.com/vi/eNKvv2sYeZo/mqdefault.jpg)
![Build with SQL + AI: From Prompt to Intelligent Apps
AI is changing how we interact with data, and you don’t need a complex setup to start building something powerful.
In this session, we’ll kick off a new SQL + AI community contest and show you exactly what it takes to participate and succeed. You’ll learn how modern AI patterns—from prompting and natural language interactions to embeddings, vector search, and retrieval-augmented generation (RAG)—connect back to SQL and real data.
We’ll focus on the practical side of building: how to take an idea, shape it with the right prompts, and turn it into something useful, whether that’s a simple interaction or a fully working intelligent app. Along the way, we’ll explore example scenarios, common patterns, and ways to think about solving real problems with SQL + AI.
This session is designed to get you ready. By the end, you’ll understand what the contest expects, how to approach it, and how to apply these concepts beyond the contest—in your own projects and real-world work!
📍This session is a part of a series. Learn more here https://developer.microsoft.com/en-us/reactor/series/S-1683/
#MSFTReactor #learnconnectbuild #DataDays
Chapter markers:
0:00 Introduction and Welcome
0:56 Speaker Introductions
2:07 SQL and AI Promptathon Overview
4:44 Data Days Resources and Certification
7:10 Agenda for Todays Session
8:05 Why Participate in the Promptathon
9:56 How to Participate in the Contest
12:35 Contest Timeline and Deadlines
13:54 Judging Criteria
16:55 AI Features in SQL Server 2025
18:02 Why Use Vector Support in SQL
20:25 The Vector Data Type
22:38 Flexibility of AI Model Integration
23:38 AI Generate Embeddings and Chunks
26:05 Vector Distance and Search Explained
31:31 Data API Builder and Security
34:26 MCP Tool Server Capabilities
37:43 Summary of GA vs. Preview Features
39:28 Demo: Getting Started with GitHub CodeSpaces
43:58 Working with Data API Builder
46:44 Exploring MCP Tools
48:32 Local Development with Dev Containers
51:00 How to Submit Your Entry
55:51 Final Thoughts and Next Steps
[eventID:27322] Build with SQL + AI: From Prompt to Intelligent Apps](https://i.ytimg.com/vi/fIDaLgkgRAo/mqdefault.jpg)
![Meet Your Claw: A Harness in Three Lines of C#
One call to AsHarnessAgent gives us the whole agentic loop. From there we add the first three abilities: a custom get_stock_price tool, the built-in web search for market news, and planning with a todo list plus plan and execute modes.
That is enough to turn a vague request like “review my watchlist and suggest something to add” into a tracked, step-by-step plan. We wire the agent to a Microsoft Foundry chat client, run it in the shared harness console, and set up the companion repo so everyone starts on the same SDK version.
📌 This event is part of a series. Learn more here: https://aka.ms/FMTA/series
🔗 Learn more by exploring the resource: https://aka.ms/AgentHarness
#MicrosoftReactor #LearnConnectBuild
[eventID:27514] Meet Your Claw: A Harness in Three Lines of C#](https://i.ytimg.com/vi/fJ1KCuhTQjU/mqdefault.jpg)
![Prove Your GitHub Copilot Skills in the flow of work
Join us to discover Microsoft Pro Badges, a new credential designed to recognize real-world achievements in the flow of work, without requiring separate tests. The first Pro Badges for GitHub Copilot, introduced in private preview at Microsoft Build, will be generally available at GitHub Universe.
In this session, youll see a live demo, get an inside look at the credential experience, and learn how to join the GitHub Copilot Verified Proficiency Insiders List for exclusive updates and early access opportunities.
https://aka.ms/AA132qba
As an added benefit, session attendees will receive an exclusive discount code for the GitHub Certified Agentic AI Developer (GH-600) Certification exam.
The discount code must be used to book the exam within 72 hours of the session, and the exam can then be scheduled for any date within the following 90 days. This offer is exclusively for session attendees and may not be shared or distributed externally.
00:00 Introduction to Microsoft Pro Badges
01:47 Why Verified Proficiency Matters
06:23 Proficiency Levels: Beginner to Expert
10:36 Beta Program, Surveys, and Learning Resources
19:21 Hands-On GitHub Skills Exercise Setup
22:10 Custom Instructions in GitHub Copilot
38:33 Building Reusable Agent Skills
50:30 Creating and Using Custom Agents
56:27 How Everyday Copilot Usage Earns Proficiency Credit
58:12 Certification Discount, GA Roadmap, and Closing Thoughts
[eventID:27475] Prove Your GitHub Copilot Skills in the flow of work](https://i.ytimg.com/vi/fUC3DEhJFGE/mqdefault.jpg)
![Get Certified SQL+AI (DP-800): Bring AI to SQL with Embeddings, Search, and RAG (APAC)
This is the session everyone’s curious about: how DP-800 connects modern AI patterns directly to your SQL solutions. We’ll break down what the exam expects around embeddings, vectors, intelligent search, and retrieval augmented generation (RAG) and translate it into practical, buildable concepts.
📍This session is a part of a series. Learn more here: https://aka.ms/datadays/SQLAI
#MSFTReactor #learnconnectbuild #DataDays #GetCertified
[eventID:27371] Get Certified SQL+AI (DP-800): Bring AI to SQL with Embeddings, Search, and RAG (APAC)](https://i.ytimg.com/vi/fYTSBb6MNyM/mqdefault.jpg)

![Using OneLake shortcuts and mirroring for cross-workspace access
This session explains how OneLake shortcuts and Mirroring in Fabric can be used to connect and share data across workspaces without making extra copies, and what that means for access control. It covers what shortcuts are, how access is evaluated across the shortcut location and the target location. We then walk through what mirroring is and how OneLake data access roles can be used to control access to mirrored data, with shortcuts to mirrored data respecting the security defined at the source mirrored item.
📍https://aka.ms/DataSecurityOverview
📌 This event is a part of a series, learn more here: https://aka.ms/OneLakeSecureAccess/series
0:00 Introduction
2:31 What is OneLake?
6:09 Data integration options
10:44 Introduction to Shortcuts
13:37 Introduction to Mirroring
18:42 Shortcut transformations
20:45 AI shortcut transformations
24:55 Demo: Stadium operations scenario
28:44 Configuring Lakehouse shortcuts
41:35 Configuring Mirrored databases
47:20 Data security and OneLake access roles
55:53 Summary and recap
#microsoftreactor #learnconnectbuild
[eventID:27041] Using OneLake shortcuts and mirroring for cross-workspace access](https://i.ytimg.com/vi/g432kE4PEFg/mqdefault.jpg)

![The Quote-to-Cash Blind Spot: Secret Scanning for Enterprise Revenue Systems
Every CPQ, CLM, and billing rollout Ive run wires Salesforce or Vlocity to DocuSign, Stripe or Zuora-type billing, and ERP through API keys and OAuth tokens sitting in DataRaptors, Integration Procedures, and deployment scripts. AppSec teams usually classify that layer as business configuration, so it never gets scanned the way application code does.
Where credentials actually live in a quote-to-cash stack: middleware scripts, integration procedures, CI/CD for CPQ deployments. Why business systems teams dont think of themselves as a scanning target, and what that costs. How GitHub Advanced Securitys secret scanning, push protection, and custom patterns close the gap, even for teams that dont see themselves as developers.
A sample repo modeling a CPQ integration script with a hardcoded OAuth token and API key. Push protection blocking a commit in real time. A custom regex pattern built for an enterprise revenue-tool token format. The alert triage view a security lead would actually work from.
#MicrosoftReactor #LearnConnectBuild
[eventID:27521] The Quote-to-Cash Blind Spot: Secret Scanning for Enterprise Revenue Systems](https://i.ytimg.com/vi/gsYbbWaowEU/mqdefault.jpg)