[PuPr] CreatorAgent: Support Semantic Models

Creator Agent is expanding support for Semantic Models in Data Agent, helping users generate semantic-model specific instructions and example queries more easily. The feature can explore Semantic Model tables, measures, columns, and recent query patterns to produce more relevant configurations while preserving transparency through visible run steps and explored queries. Users remain in control by reviewing, refining, and committing generated configurations before they update their Data Agent setup.

[PuPr] Data Agent to Upgrade Ontology Integration

Microsoft Fabric Data Agent is improving its Ontology experience so customers can build agents that deliver more accurate, reliable, and verifiable answers grounded in their business concepts and relationships. With this release, customers should see better support for complex questions, clearer visibility into answer generation, stronger customization and scoping, and improved confidence as ontologies evolve over time.

[PuPr] Creator Agent Supports Files as Context

Creator Agent file support lets users upload existing text-based business context such as documentation, glossaries, process guides, and data dictionaries, so Creator Agent can use it to generate more complete and accurate Data Agent configurations. This reduces repetitive setup work by allowing users to reuse knowledge that already exists in files, rather than manually recreating that context through long chat interactions.

Open Fabric content in a new tab with right-click

Expands right-click support across key Fabric surfaces, enabling users to quickly open content in a new browser tab without navigating away.

Organize multitasking tabs in groups

Stay focused when working across many open items by grouping tabs. Create named groups, organize related work, and collapse or expand them as needed.

Enable stream processing and real-time analytics on Mirror DB change feeds (GA)

### Mirrored Database Change Feed ConnectorThe Mirrored Database Change Feed Connector streams Delta Change Feed (CDF) updates from Fabric Mirrored Databases directly into Eventstreams. Build real-time analytics, dashboards, alerts, and automated actions using familiar no-code and SQL experiences in Eventstream, Eventhouse & Activator. Supports all Mirrored Database sources that expose Delta Change Feeds through Mirror Database Extended Capabilities.

DeltaFlow: Seamless stream processing of Database CDC Feeds (GA)

DeltaFlow is an enhancement to Eventstreams that simplifies real-time analytics on Change Data Capture (CDC) feeds. By automatically transforming raw Debezium change events into structured tables that mirror the source schema, DeltaFlow enables developers and analysts to work with change data as if it were regular tables--eliminating the need to interpret complex CDC semantics. This streamlined approach accelerates the development of operational dashboards, event-driven applications, and real-time analytics solutions, allowing users to extract insights, build AI-powered workflows, and react instantly to changes in their data sources with minimal manual setup and maintenance.

Large message support ( 20 MB) for Fabric Event streams

### Large Message Support in Fabric Eventstreams #### Overview Large Message Support enables Fabric Eventstreams to ingest, process, and deliver event payloads up to **20 MB** per message. This removes the need for customers to split, chunk, or pre-process large events before streaming them through Eventstreams. #### Problem Today, Eventstreams enforces a default message size limit of 1 MB. Customers with larger payloads -- such as batched IoT telemetry, enriched event schemas, media metadata, or complex domain events -- must implement workarounds to stay within this constraint. These workarounds add pipeline complexity, increase failure surface, and break downstream schema consistency. #### What's changing With this feature, customers can opt in to a configurable message size ceiling (up to 20 MB) on their Eventstreams. Key aspects: - **Opt-in by design** -- existing Eventstreams are unaffected unless the customer explicitly enables a higher limit. - **Configurable ceiling** -- customers choose the maximum message size aligned to their workload, rather than a one-size-fits-all default. - **No pipeline redesign required** -- sources, transformations, and destinations continue to work as expected. Larger messages flow through the same Eventstream canvas.

Lakehouse Delta Table Change Feed Connector (Preview)

### Lakehouse Delta Table Change Feed ConnectorThe Lakehouse Delta Table Change Feed Connector reads Delta Change Feed records directly from change-enabled Delta tables and streams them into Eventstreams, enabling incremental processing, real-time analytics & dashboards, and AI-powered applications using Eventstream no-code/SQL operators, Eventhouse Real-Time Dashboards and Activator.

Easily use Spark Structured Streaming with Fabric Eventstreams (GA)

### Eventstream & Spark IntegrationFabric Eventstream & Spark Integration brinsg together Real-Time Intelligence and Fabric Spark in a single end-to-end experience. It enables data engineers and data scientists to seamlessly discover, consume, and process Eventstream data using Fabric Notebooks and Spark jobs to build real-time analytics, AI applications, dashboards, and event-driven actions.

Git Integration - Allow contributor and above users to switch branch

Today, when a Fabric workspace is connected to a Git repository, only workspace admins can switch the selected branch. With this change, Contributor and above users can also switch branches. Additionally, Contributor and above users will be able to branch out to existing workspaces that have this setting enabled, even if they aren't admins on the target workspace. This new capability will be available via Fabric UI and Fabric REST API.

Git Integration provider - GitHub Enterprise Cloud with data residency support

Ability to connect Fabric workspace to repositories which reside on GitHub Enterprise Cloud with data residency

Salesforce CDC source connector

Salesforce change data capture events source connector

Bring your own Event Hub to Fabric Eventstreams (Mounted eventstreams)

Eventstream connector custom CA and mTLS support - GA

Live delivery insights in Eventstream

Eventstream Observability Log support

Enable CopyJob sources in Eventstream (PuPr) + RTH integration

Azure Data Explorer source connector - GA

Salesforce platform events source connector

Salesforce platform events source connector

Enable Eventstreams as CopyJob source & destination

ServiceNow zero-copy integration

ServiceNow Workflow Data Hub adds a zero-copy connector to Microsoft OneLake. Workflow Data Fabric is an enhanced data integration layer that unifies business and technology data across the enterprise, powering all workflows and AI agents with real-time, secure access to data from any source. With this integration, customers can access tables in OneLake using the Table APIs powered by Apache Iceberg.

Intuigence AI

Synthetic AI engineers for Industrial Operations that help users automate routine tasks so human engineers can bring their ingenuity to creative and complex challenges.Its industrial semantics, ontologies, physics aware models and purpose built agents supply the understanding that turns raw data into industrial grade operational intelligence using RTI on Fabric as the backbone. Eventstreams capture and transform data at massive scale, eventhouse delivers fast and flexible analytics and OneLake ensures that context is unified and discoverable.

Outbound access protection for Eventstream

Outbound Access Protection (OAP) in Microsoft Fabric restricts outbound connectivity from a workspace to external destinations unless explicitly allowed by administrators, helping prevent data exfiltration and enforce security policies.