Microsoft Fabric is getting new tools and capabilities to streamline artificial intelligence application development, enhance data integration and improve security in a series of announcements at Microsoft Corp.’s Ignite conference today.

Fabric is an integrated analytics platform that unifies data integration, engineering, warehousing, science and business intelligence on top of a unified data lake. The updates are aimed at improving developer, analyst and data scientist productivity and simplifying access to data through a catalog.

New Fabric Databases, now in preview, are intended to provide simpler database provisioning with autonomous functions. The first available volume, SQL Databases in Fabric, allows data to be automatically replicated to Fabric’s integrated OneLake data lake for immediate use by Fabric’s analytical tools. A feature native vector search combined with Azure AI embedding models supports advanced artificial intelligence functions like retrieval-augmented generation.

Native vector search is a technique in which data is represented as high-dimensional vectors, and search is performed by finding items whose vectors are closest to a query vector, rather than necessarily being an exact match. Vector data is a critical element in machine learning model training and semantic search.

The auto-optimizing and auto-scaling databases provide faster application performance through intelligent indexing, Microsoft said. Developers can leverage a Fabric Copilot to translate natural language queries into SQL as well as complete code and deliver in-line explanations.

Integration with GitHub and Visual Studio Code, a lightweight, open-source code editor developed by Microsoft, support continuous integration and deployment and simplify database updates and source control. A common data estate ensures consistent governance policies.

Easier generative AI

Microsoft is also expanding AI capabilities within Fabric to make generative AI more accessible. AI Functions in Fabric is a new feature that allows users to perform text-related tasks such as summarization, translation and sentiment analysis with minimal coding. Simplified application programming interfaces reduce AI enrichment complexity. Azure AI Agent Service Integration enables simplified connection to enterprise data sources, including SharePoint, to automate tasks.

A new Open Mirroring capability, now in preview, allows applications and data providers to more easily bring external data estates into OneLake. Mirroring simplifies data replication by creating an exact copy of data from one system to another in a way that ensures that both the primary and secondary systems have identical data at any given point in time.

The OneLake Catalog, now generally available, provides a unified platform for exploring, managing and governing Microsoft Fabric data. It provides a central interface for discovering data, including filters for domains, owners and tags. New features coming next year will help with data quality, labeling and compliance with actionable recommendations.

Image: Microsoft