Once again there were a number of Microsoft Build announcements related to data and AI, and some were very impressive. Below are my favorites.

Everything announced at Build can be found in the Microsoft Build 2025 Book of News.

Here you go:

Cosmos DB in Fabric (preview): First there was SQL Database in Fabric, and now Cosmos DB. You can now store semi-structured NoSQL data in Cosmos DB in Fabric, alongside your relational data in SQL databases, enabling a unified data platform for your applications. This further positions Fabric as a complete data platform to handle all your organizational needs, from operational to analytics and BI. Check out this video to see it in action. You can try Cosmos DB in Fabric today or learn more by reading the Cosmos DB in Fabric blog.

Digital twin builder in Fabric (preview): A new capability designed to help organizations bridge their physical and digital worlds to create an AI ready foundation for their operations. It simplifies the creation and management of data-driven digital twins at scale, enabling users to understand system-wide connections and create unparalleled insights and operational efficiency. More info. Here is a video demo.

Microsoft 365 Copilot Tuning (available via Early Access Program): This is huge! Offers customers a new way to unlock the value of fine-tuning for organizations without the cost and complexity of other solutions.  Now, makers can use the low-code tooling in Microsoft Copilot Studio to take advantage of highly automated fine-tuning “recipes” that can use your enterprise data to train models to assist with domain-specific tasks.  Once training is finished, agents, built in Agent Builder with just a few clicks, can then tap into these fine-tuned, task-specific models and easily integrate them into the Microsoft 365 apps. More info and video.

Copilot Studio multi-agent orchestration: Also huge! Enables agents to exchange data, collaborate on tasks, and divide their work based on each agent’s expertise. For example, multiple agents can collaborate across HR, IT, and marketing to help onboard a new employee. See this Video and more info.

Warehouse Snapshots in Microsoft Fabric (preview): A new capability to provide a stable, read-only view of your data warehouse at a specific point in time. With Warehouse Snapshots, you can confidently support analytics, reporting, and historical analysis without worrying about the volatility of live data updates. More info.

Materialized Lake Views in Fabric (preview soon): Allows you to build declarative data pipelines using SQL, complete with built-in data quality rules and automatic monitoring of data transformations. In essence, an MLV is a persisted, continuously updated view of your data that simplifies how you implement multi-stage Lakehouse processing, commonly referred to as medallion architecture. More info.

Mirroring for SQL Server in Microsoft Fabric (preview): For all in-market versions of SQL Server from SQL Server 2016 to SQL Server 2025. More info.

Microsoft Fabric SKU estimator: Predict your capacity needs by entering details about how your team plans to use Fabric. Use it.

Fabric roadmap tool: Brings the upcoming features all together in one place, with a cleaner interface, real-time updates, and direct integration with the internal planning tool used by the Fabric team. See it.

Shortcut transformations (preview): Introduces the ability to transform data as it’s shortcut into Fabric including changing the data format into Delta tables or applying AI transformations to unstructured data—such as summarizing text, translating content, or classifying documents. More info.

Fabric data agent integration with Microsoft Copilot Studio (preview): 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, Microsoft is bringing these data agents into Copilot Studio. Once you connect your Fabric data agent to your custom agent in Copilot Studio, it 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.  Check out the video.

Fabric data agent with Copilot in Power BI (preview soon): Previously, you may have access to multiple resources, but finding the right data to answer specific questions can be challenging. Copilot was limited to the right pane of a single report, allowing questions only about that open report. This new standalone Copilot in Power BI addresses this by streamlining the process. This full-screen Copilot experience, easily accessible from the left navigation of Fabric, can help you find and ask questions about any data you have access to. It serves two purposes. First, you can ask Copilot to find Power BI reports, semantic models, apps (coming soon), and Fabric data agents that you have access to. It will match across many different factors to quickly find the most relevant items (if you already know which resource to use, you can manually add it to the Copilot session and interact with it directly for more relevant results). Second, it allows you to ask natural language questions and receive accurate, relevant answers from your available Fabric resources. For example, using Copilot’s summary capabilities allows you to quickly identify the most interesting data within your report. Often, reports can become quite complex, and you can easily spend 30 minutes to a couple of hours combing through all the details. However, Copilot can help you sift through the report, giving you easy to digest overviews of your data. Just ask Copilot to ‘Tell me about trends in sales’ or more specific topics, like ‘What should I know about bike sales in Washington?’. You may also have a business question that cannot be answered by the existing report. Typically, what you’d have to do in these cases is work with an analyst to get a new visual added to the report, which can take time, delaying your ability to make a data driven decision. This can be especially cumbersome if it’s not something you’d want to track long term. Also, with the standalone Copilot experience, you can easily ask questions on a semantic model, allowing you to quickly find answers to ad hoc questions. Finally, you can also ask questions against your Fabric data agents.  Check out the video and more info and more info.

Prepare your data for AI: Power BI is introducing new capabilities to help you get your data ready for natural language experiences with Copilot. The first step to get the best results from Power BI Copilot is always to have a semantic model that follows best practices. However, models often require more assistance to be fully prepared for the optimal interactions with AI. Now available in Power BI Desktop is tooling features to help you prepare your data for AI (coming soon for the Power BI service). These features allow you to provide more context about your model, help guide Copilot to the right data in the model and help increase the quality of Copilot output. More info.

Dataflow Gen2 parameterization (preview): Leveraging query parameters while authoring Dataflows Gen2 has been possible for a long time, however, it was not possible to override the parameter values when refreshing the dataflow. The ability to pass values from a pipeline into a Dataflow parameter for refresh has been one of the top ideas in the Fabric ideas portal since Dataflow Gen2 was released.

SQL Server 2025 (public preview): Microsoft announced the public preview of SQL Server 2025 at Build, positioning it as the AI-ready enterprise database. This release brings native AI integration into the SQL engine, allowing developers to build intelligent applications using familiar T-SQL syntax. Key features include built-in vector search for semantic queries, enhanced support for JSON and regular expressions, and compatibility with popular AI frameworks like LangChain and Semantic Kernel. It also introduces an open-source Python driver and GitHub Copilot integration within SQL Server Management Studio (SSMS) 21. Security and performance improvements include support for Microsoft Entra managed identities, optimized locking, and intelligent query processing. SQL Server 2025 also connects seamlessly with Microsoft Fabric for real-time analytics and supports hybrid deployments via Azure Arc. More info.

More info:

Powering the next AI frontier with Microsoft Fabric and the Azure data portfolio

Get to insights faster with SaaS databases and “chat with your data”

Introducing Microsoft 365 Copilot Tuning, multi-agent orchestration, and more from Microsoft Build 2025

Fabric May 2025 Feature Summary