The emergence of an entirely new type of organization—reconstructed with AI—was uncovered in Microsoft’s latest annual Work Trend Index report. These frontier firms are building agent assistants, creating hybrid human and agent teams, and even establishing entire teams of agents directed by humans. As more organizations begin their journey to this next frontier, agents will begin to operate across every individual and team, with organization-wide context, automating tasks and providing humans with timely, contextual insights.
Developers will be at the heart of this agentic web. But powering these agents will require more than just AI models. Developers will need to bring together every type of data an organization produces; not just analytical, but transactional and operational, in both structured and unstructured forms. Take The Estée Lauder Companies, for example, the global beauty company trained an agent on their structured consumer insight data to provide their teams with actionable intelligence instantly. Whereas dentsu, a global marketing firm, trained a suite of agents on their unstructured HR and compliance data to quickly answer employee questions.
Developers will need the right tools at their disposal to bring together all this data, prepare it for AI, and use it to train the agents that will soon become our digital teammates. Whether you need to work with analytical, operational, or transactional data, we are making sure you have the tools you need. In fact, Microsoft is recognized as a leader across the 2024 Gartner? Magic Quadrant??for Data Integration Tools, Cloud Database Management Systems, Cloud AI Developer Services, and Analytics and Business Intelligence Platforms.
In this blog, I’ll cover the latest slate of announcements across Microsoft Fabric and the entire Azure Data portfolio, designed to empower every developer on the planet to do more with data:
- Fabric announcements: Cosmos DB (NoSQL) in Fabric, a new experience called digital twin builder, and new ways to use Copilot to chat with your data.
- Azure Data announcements: SQL Server 2025, PostgreSQL in VS Code with GitHub Copilot, AI Foundry integration with Azure Cosmos DB and Azure Databricks, and SAP integration with Azure Databricks.
Empowering app developers with new tools in Fabric
Microsoft Fabric is our AI-powered platform designed to bring all your teams and data together to accomplish any data project. With Fabric, our goal is to converge all the data services you need into a unified, open, and extensible platform, so you no longer have to manually stitch together disconnected services. This vision of a converged data platform has already resonated with more than 21,000 customers—including over 70 percent of the Fortune 500—who are using Fabric today. And with more than 50 percent of our customers using more than three workloads, most see the true value of Fabric as a do-it-all data platform.
By converging our industry-leading tools in a single software-as-a-service (SaaS) platform, we can also help developers accelerate their projects. No matter what your data project requires, whether it’s managing databases, accessing real-time data, or training machine learning models, Fabric has the tools you need, which work together seamlessly out of the box. Since Fabric is SaaS, you can get started instantly without the complexity of infrastructure and configuration settings you typically find in data platforms. Continuous integration and continuous delivery (CI/CD) capabilities are woven into the platform along with direct integration to the tools developers use every day, like GitHub and VS Code.
And importantly, with Fabric, developers can access and analyze any type of data. Microsoft OneLake, Fabric’s open data lake, can connect to structured and unstructured data across any cloud or format. Fabric Real-Time Intelligence can support your data in motion. And with our most recent announcement of Fabric Databases, we can help you bring your transactional scenarios to Fabric. Today, we are enhancing our support for your operational, semi-structured data with the addition of Cosmos DB in Fabric.
Accelerate app development with Cosmos DB in Fabric
Six months ago we announced Fabric Databases, a new class of SaaS databases built directly into Fabric that are easy to deploy and manage and instantly available to help developers streamline application development. We started with SQL databases in Fabric, which you can provision in seconds for your structured data and are highly available and secure by default.
However, as agents take on increasingly complex tasks, the ability to bring together semi-structured data like text, documents, emails, and graphs will prove critical for AI training. Which is why we are thrilled to announce we are expanding Fabric Databases to handle semi-structured data with the preview of Cosmos DB in Fabric. According to a recent Bloomberg CIO study, Azure Cosmos DB was the top choice for building generative AI applications.1 In fact, OpenAI chose Azure Cosmos DB as the database to support the trillions of database transactions and petabytes of data needed to support the 500 million users who use ChatGPT weekly.2
We are taking this industry-leading database technology and bringing it to Fabric. With Cosmos DB in Fabric, developers can deploy a high-performance database with just a few clicks while still experiencing enterprise-grade dynamic scalability and 99.999% reliability. With support for both SQL and NoSQL models, developers now have the flexibility to build AI applications grounded in operational, transactional, and analytical data. Best of all, Cosmos DB data is instantly available in OneLake for analytics like near real-time sentiment analysis for chat applications. Check out these new capabilities in action:
You can try Cosmos DB in Fabric today or learn more by reading the Cosmos DB in Fabric blog.
Bring the physical world to the digital world with digital twin builder in Fabric
With your data accessible in Fabric, we’re also bringing you more tools to analyze and uncover insights from your data. One of the most exciting new tools is called digital twin builder, now in preview. Digital twin builder is a powerful new capability that enables organizations to create, manage, and visualize virtual replicas of physical and logical entities at scale. Built in Fabric Real-Time Intelligence, digital twin builder provides a simpler, no-code or low-code way to build and manage virtual representations of real-world objects and processes.
With digital twin builder, you can connect, map, and manage virtual replicas of physical and logical entities, whether they are physical assets like machinery, logical entities like customers, or dynamic processes like manufacturing and logistics. You can use this digital twin to enhance deep analytics, perform what-if analysis, and automation your processes. If you want to learn more, you can read more about digital twin builder in Fabric here, and you can also start your 60-day Microsoft Fabric free trial as well. Check out the following video of digital twin builder in action:
Empower everyone to ask questions about their data in Power BI and Microsoft Copilot Studio
Insights are only impactful when they reach those who can use them to inform decisions. That’s why making data more accessible through chat experiences is one of the fastest growing AI experiences. A trained agent can enable teams to simply ask questions about their data, providing a more accessible and interactive way to uncover insights.
Chat with your data through Copilot in Power BI
Now, we are thrilled to announce a new standalone, full-screen Copilot experience in Power BI which allows everyone to chat with the data they have access to. With over 30M monthly active Power BI users and embedded experiences in the apps we use every day, like Microsoft 365, Power BI has become the default source of data and insights for many business users. In the coming weeks, users will be able to open Copilot on their home screen and ask broader questions about their data. Copilot will automatically search across multiple reports and semantic models to intelligently retrieve the most relevant data you have access to and answer your questions. You can even access this Copilot in Power BI experience directly from Microsoft Teams, so you don’t have to break up your normal flow of work to get answers. Check out the following demo to see how this works:
I’m excited to share that we are working to extend Copilot in Power BI capabilities into Microsoft 365 Copilot with Power BI agent, allowing users to find content, ask questions, and visually explore and analyze data without leaving Microsoft 365 Copilot.
Connect Fabric data agents to Microsoft Copilot Studio
In Fabric, you can create data agents—virtual business analysts which enable you to engage in natural language conversations about your data in OneLake. Building on this capability, we’re excited to bring these data agents into Copilot Studio. Once added to your agent in Copilot Studio, they can be deployed across Microsoft Teams and Microsoft 365 Copilot to reason over complex datasets, get insights directly from data in OneLake (respecting data access permissions), and take action. Your agents can even automate tasks like sending emails or triggering workflows, making it easier for users to interact with enterprise data and make data-driven actions in context.
Read more about the new chat with your data experience in Power BI and stay tuned for when the Fabric data agents integration with Copilot Studio becomes available in the coming weeks.
Transform your data into a competitive advantage with Azure Data
For decades, databases have been the backbone for managing enterprise information—enabling efficient storage, retrieval, and analysis that power the world’s applications. And now, as we shift into the AI era, databases are becoming even more important to support our increasingly AI-powered applications.
The ODP Corporation, which includes Office Depot, uses Azure Cosmos DB together with Azure AI services to serve HR data to employees in real-time, replacing turnarounds that used to take 24 hours.
We use Azure Cosmos DB to store HR data at the profile level. It’s the glue between our back end and front end—everyone who logs in gets a profile, and the model reads and writes to it in real time to maintain context.
—Mick Feller, Distinguished Engineer, The ODP Corporation
The ODP story is just one of many, with customers like Docusign, BNY Mellon, and Mondra similarly using Azure Databases to fuel their AI projects.
We are working hard to further enrich our Database offerings so they keep up with AI model innovation and continue fueling your AI applications. Today, we’re excited to unveil major announcements across both our industry-leading SQL and open-source database offerings, designed to assist developers in crafting intelligent applications.
Build AI apps securely from ground to cloud with SQL Server 2025
For over 35 years, SQL Server has been an industry leader in providing secure, high-performance data management. Now, we are thrilled to announce SQL Server 2025 is officially in public preview.
Our newest version of SQL Server is purpose-built to securely support your AI applications, transforming into a vector database in its own right. SQL Server uses built-in filtering capabilities along with a vector search, with flexible interfaces for AI models deployed locally or in the cloud, simplified workflows with integrated vector embedding, and support for popular frameworks like LangChain, Semantic Kernel, and Entity Framework Core.
SQL Server 2025 has already resonated well with more than 3,400 applicants for our private preview program, with full adoption coming twice as fast as SQL Server 2022. You can also get started through frictionless integration with Fabric and Azure Arc, with support for real-time data replication in Fabric through mirroring. See SQL Server 2025 in action and learn how you can get started today.
Bring together Azure PostgreSQL, VS Code, and GitHub Copilot
Open source databases are playing an increasingly more critical role in shaping the future of intelligent applications, with PostgreSQL being the most popular according to Stack Overflow’s 2024 Developer Survey.3 Microsoft is committed to supporting open source and making the developer experience better than ever. Notably, Microsoft has the most Postgres committers and contributions to the open-source community amongst all the hyperscalers.
Today, we’re announcing the preview of the new and improved PostgreSQL extension for VS Code with GitHub Copilot built in—designed to simplify workflows and boost productivity through AI-powered assistance. The PostgreSQL extension allows developers to manage databases directly within VS Code, whether they’re working in Azure, Docker containers, or on-premises environments. Now, the GitHub Copilot integration brings natural language capabilities to database development, helping developers design schemas, write optimized queries, and troubleshoot issues with expert-level guidance, right from their editor.
We are also announcing the general availability of DiskANN on Azure Database for PostgreSQL, one of the fastest vector indexing algorithms on the market. Developers can leverage DiskANN to build high-performance, low-latency, and scalable generative AI applications that surpass pgvector index types.
Vector search has been foundational to building generative AI applications, but it has limitations when it comes to understanding some semantic relationships between enterprise data. We’re making it easier for teams developing intelligent applications and agents to unlock deeper insights from their operational data by introducing generative AI-powered reasoning in PostgreSQL. These new semantic operators in Azure Database for PostgreSQL, now in preview, leverage LLMs directly within the database to help you unlock deeper relational context from data.
Connect to Azure AI Foundry directly from Azure Cosmos DB and Azure Databricks
For developers designing and customizing AI apps and agents, we’re announcing that you can now use Azure Cosmos DB accounts to power AI solutions in Azure AI Foundry.
Customers can now securely store and manage the conversation threads between users and AI agents in their Azure Cosmos DB accounts, using the Azure AI Foundry SDK. This enables agents to recall the content of previous thread conversations and messages and pick up conversations where they left off. Threads storage is now generally available.
Coming soon, developers will be able to use the data stored in their Azure Cosmos DB accounts to power AI solutions in Azure AI Foundry. Customers can soon connect and access their Azure Cosmos DB data using Azure AI Foundry in application code, and Azure Cosmos DB will be the first Azure database able to power agents and models in Azure AI Foundry with real-time, operational data.
We’re also announcing the public preview of Azure AI Foundry connection for Azure Databricks. This connection centers on two key scenarios: enabling Foundry Agents to use AI/BI Genie and run Azure Databricks Jobs. These capabilities can enhance knowledge retrieval and expand how Foundry Agents deliver contextual answers grounded in enterprise data, where much of the world’s data resides. And, if you want to learn more about all of the Azure AI Foundry announcements, you can read their blog here.
Expanding our Databricks and SAP partnership
We are thrilled to deepen our longstanding partnership with SAP and Databricks, supporting SAP’s latest innovations with the launch of SAP Business Data Cloud (BDC) and SAP Databricks on Azure, expected in Q3 2025. SAP Databricks is a core component of SAP Business Data Cloud (BDC) that enables users to leverage analytics capabilities of Databricks tailored to SAP workloads. Now developers can innovate on top of the core business data running in their SAP applications. When running on Azure, SAP Databricks will be powered by Microsoft’s first party Azure Databricks service. This integration allows SAP Databricks to benefit from robust scalability, performance, security, and the Azure integrations customers expect from native Azure services.
Connect your Azure Databricks data to Fabric
You can use mirroring to access your Azure Databricks Unity Catalog tables in OneLake, currently in preview, and keep them in sync in near real-time. Simply add your Azure Databricks workspace URL, select the catalog, and Fabric creates a shortcut for every table in the selected catalog. Learn more by watching this video or by viewing the documentation.
Moving forward with data and AI innovations in Azure
With these announcements, we are evolving our data offerings alongside our AI innovations to keep pace in this new era. These new innovations are designed to help developers break down silos and integrate their data into AI applications, no matter the type of data or where it lives within an organization.
Watch these announcements in action at Microsoft Build
Join us at Microsoft Build?from May 19 to 22, 2025, to see all of these announcements in action across the following sessions:
- BRK206: Microsoft Fabric for Developers: Build scalable data and AI solutions.
- BRK204: What’s new in Microsoft Databases: Empowering AI-Driven App Dev.
- BRK202: Scale and secure MongoDB-compatible apps with Azure Cosmos DB.
- BRK203: Get faster LLM responses and low app latency with Azure Managed Redis.
- BRK205: What’s coming in Fabric Automation and CI/CD.
- BRK207: SQL Server 2025: The Database Developer Reimagined.
- BRK208: Building AI agents for actionable insights with data in Fabric.
- BRK209: Building real solutions with Real-Time Intelligence in Fabric.
- BRK210: Build AI apps and unlock the power of your data with Azure Databricks.
- BRK211: Building advanced agentic apps with PostgreSQL on Azure.
- BRK212: Design scalable data layers for multi-tenant apps with Azure Cosmos DB.
- BRK213: Enable advanced AI scenarios with Unified Data Estates in Microsoft Fabric.
Sign up now for our upcoming security webinars:
- May 28: Ask the Experts—Securing your data in Microsoft Fabric: A webinar where experts from across Fabric security will join to answer all your questions live.
- June 16: Connect Securely to Your Most Sensitive Data in Fabric: A webinar with Rick Xu who will demo the latest network security features in Fabric.
References
2 Forbes, ChatGPT Hits 1 Billion Users? ‘Doubled In Just Weeks’ Says OpenAI CEO