👤 Who is this for?

All Roles — Stay up to date with the latest Microsoft Fabric announcements, feature releases, and blog posts across all workloads.

Guide Updates

What's New in This Guide

Recent additions and improvements to the Microsoft Fabric Best Practices Guide.

🆕 Frameworks: WAF + CAF

New Frameworks page covering the Well-Architected Framework for Fabric (5 pillars: Reliability, Security, Cost Optimization, Operational Excellence, Performance Efficiency) and the Cloud Adoption Framework (4-step organizational adoption strategy). Based on official Microsoft docs published May 2026.

May 2026

🆕 Synapse to Fabric Migration

Comprehensive Synapse Migration guide covering the evolution from Azure Synapse to Fabric, migration paths for Data Warehouse and Spark workloads, official Migration Assistant tools, data type mapping, migration patterns, and a decision framework.

May 2026

🔄 Site-Wide Content Audit

Updated terminology across all pages: "Real-Time Analytics" → Real-Time Intelligence, "Azure AD" → Microsoft Entra ID, "KQL Database" → Eventhouse. Refreshed P-SKU references to clarify legacy status. Added disclaimers on CU smoothing and overage pricing. Cross-referenced operations migration section with new dedicated page.

May 2026

🆕 CI/CD & Git Integration

New CI/CD page covering Git integration setup (Azure DevOps & GitHub), deployment pipelines, branching strategies, REST API automation, GitHub Actions examples, and environment promotion patterns.

May 2026

🆕 Monitoring & Observability

New Monitoring page covering the Capacity Metrics App, Workspace Monitoring, OneLake Diagnostics, Fabric Activator alerts, FUAM (admin monitoring), Azure Monitor / Log Analytics integration, and audit log pipeline patterns.

May 2026

🆕 AI & Copilot in Fabric

New AI & Copilot page covering Copilot across all workloads (Power BI, Notebooks, Pipelines, SQL), Fabric Data Agents, AI Skill for natural language querying, and AI governance best practices.

May 2026

🆕 Multi-Cloud & Hybrid Patterns

New Multi-Cloud page covering AWS S3/Redshift, GCP BigQuery, Snowflake, and Databricks integration patterns, on-premises hybrid architectures, landing zone design, and a decision matrix for choosing integration approaches.

May 2026

🆕 Fabric Databases (SQL DB in Fabric)

New Fabric Databases page covering SQL Database in Fabric, Database Hub, autonomous tuning, Copilot for SQL, and a decision matrix comparing Fabric DB vs. Data Warehouse vs. Lakehouse.

May 2026

🆕 Dedicated Security Page

Comprehensive defense-in-depth security guide with 8 interactive layers covering identity, workspace, item permissions, data security (RLS/CLS/OLS/DDM), OneLake security, information protection, network security, and monitoring & audit.

April 2026

Governance Page Refocused

The Governance page has been refocused to cover Purview integration, domains, endorsement, and workspace organization — with security content moved to the dedicated Security page for deeper coverage.

April 2026
Build 2026

Microsoft Build 2026 Announcements

Key Fabric and Power BI announcements from Microsoft Build 2026 — the agentic era of data and analytics.

🚀 Microsoft Build 2026 — June 2026

Build 2026 marks Fabric's evolution from a data platform to a full enterprise intelligence platform. The theme: context — not just model capability — is the bottleneck for AI/agent adoption. Fabric fills that gap by sharing business context across all agents and apps. The new Microsoft IQ context platform (GA) unifies Work IQ, Fabric IQ, Foundry IQ, and Web IQ to ground agents in enterprise knowledge.

Microsoft Fabric

Rayfin — Backend-as-a-Service for Fabric

A fully managed, open-source Backend-as-a-Service (BaaS) platform that enables both developers and AI agents to define and deploy complete application backends onto Microsoft Fabric in minutes. Think Supabase/Firebase — but with enterprise governance baked in.

  • Code-first: TypeScript decorators define data models, auth, APIs, access policies, and storage
  • 3-command deployment: npm create @microsoft/rayfin → npx rayfin up → production
  • GraphQL APIs out-of-the-box — schema-driven, auto-generated from your models
  • OneLake-native: All data lands directly in OneLake — instantly available for Power BI, notebooks, and agents
  • Enterprise governance inherited: Compliance, RBAC, and security from Fabric environment — no manual wiring
  • AI-agent aware: GitHub Copilot and coding agents can generate/modify backends directly
  • Replit partnership: Prototype in Replit, deploy to Fabric for enterprise-grade production
  • Open-source: SDK & CLI on GitHub
Getting Started with Rayfin
# Scaffold a new Rayfin project
npm create @microsoft/rayfin@latest my-app --workspace my-fabric-workspace

# Deploy to Fabric (provisions DB, auth, APIs)
npx rayfin up

Fabric IQ — Generally Available

The shared semantic foundation of Fabric is now GA. Fabric IQ provides business context for agents, applications, and analytics workflows — an ontology layer that grounds AI in business meaning, not just raw schemas.

  • Ontology-based grounding: Agents reason using business concepts (e.g., "revenue," "customer churn") not table names
  • Part of Microsoft IQ: Works alongside Work IQ (M365 signals), Foundry IQ (institutional knowledge), and Web IQ (real-time web data)
  • MCP-compatible: Model Context Protocol enables seamless agent-to-agent orchestration
  • Web IQ: New model-agnostic web grounding — ~2.5× faster relevant information vs. competitors
  • Work IQ APIs (June 16): Programmatic access to people, emails, documents, and meeting signals
  • Core layer for the unified agentic platform — enables multi-agent systems with shared semantics

Unified Agentic Platform

Fabric positions itself as the end-to-end platform for building, operating, and governing autonomous AI systems. The focus shifts from writing individual apps to orchestrating entire multi-agent systems with Fabric as the orchestrator of context and intelligence.

  • Data Agents (GA): Virtual analysts aligned to specific domains — generate insights and take actions across OneLake, lakehouses, semantic models, and warehouses
  • Operations Agents: Monitor real-time data, detect patterns, and take proactive action without human intervention
  • Cross-platform: Agents usable across Fabric, Microsoft Foundry, Copilot Studio, and M365 Copilot
  • Graph integration: Native graph database with GQL for modeling entity relationships at scale
  • Database Hub (Early Access): Unified management for Azure SQL, Cosmos DB, PostgreSQL, MySQL, SQL Server (Arc), and Fabric Databases — Copilot-powered insights and agent-assisted recommendations
  • New Database Savings Plan — up to 35% cost savings

Real-Time Intelligence & Architecture Evolution

Fabric transitions to a full enterprise intelligence platform — emphasizing contextual intelligence, real-time analytics, and embedded AI/agent operations.

  • OneLake as unified foundation: Non-lock-in data layer accessible by all Fabric features, agents, and external tools
  • Contextual intelligence: The bottleneck is no longer model capability but context — Fabric provides the missing enterprise context layer
  • MAI Model Family: New in-house reasoning models (MAI-Thinking-1 with 256K-token context window) available in Foundry
  • Microsoft Scout: New personal agent for work based on OpenClaw and WorkIQ (preview)
  • New governance and real-time data management tools — remove silos while maintaining compliance

Power BI

Agentic Analytics & Copilot

Power BI enters the Agentic Era — the analytics platform now natively supports AI agents that can act, execute, and automate across your data stack. Copilot handles multi-step instructions, generates and refines visuals, and enables fully conversational analytics on desktop and mobile.

  • DAX generation & explanations: Natural language → DAX measures, columns, and queries (85–90% accuracy for simple, 60–70% for complex)
  • Report generation: Create complete reports, visualizations, and narrative summaries from datasets
  • Conversational analytics: Ask questions in natural language, get direct answers — now fully available on mobile
  • Multi-step instructions: Chain commands like "filter to last quarter, break down by region, highlight anomalies"
  • Modeling MCP Server: Enables bulk/agentic changes (renaming, translation, refactoring) across Power BI models
  • Licensing: Requires Fabric capacity (F64+) or Premium Per User (PPU) — no separate Copilot license
  • Q&A deprecated: Retirement planned December 2026 — Copilot replaces Q&A functionality

DAX User-Defined Functions (GA)

Define custom, reusable DAX functions across different Power BI models and reports. Copilot can now invoke your UDFs directly — describe usage in AI instructions and Copilot calls them for sophisticated analyses on demand.

  • Write once, reuse everywhere: Eliminates copy-paste DAX patterns across models
  • Copilot-aware: Describe your UDF in AI Instructions — Copilot invokes it when answering complex prompts
  • Enterprise consistency: Enforce calculation standards across all reports and teams
  • Improve readability — abstract complex logic behind named functions
DAX UDF Example — ABC Analysis
DEFINE FUNCTION ABCAnalysis (
    AUpperBound AS INT,
    BUpperBound AS INT,
    AnalysisDate AS DATETIME
) =
    // Custom classification logic here
    VAR CumulativePct = ...
    RETURN
        SWITCH(TRUE(),
            CumulativePct <= AUpperBound, "A",
            CumulativePct <= BUpperBound, "B",
            "C"
        )

-- Copilot prompt: "Do an ABC analysis on Transactions for Q1"
-- Copilot automatically calls your ABCAnalysis UDF!

Direct Lake & Semantic Model Improvements

Enhanced capabilities for analyzing data directly from OneLake without full dataset refreshes. The semantic model becomes the key interface between business data and AI.

  • Import-class performance: No data movement — queries run directly against OneLake Delta tables
  • Outbound access protection: New security controls for semantic models
  • Semantic model settings pane: Simplified management experience
  • Fabric IQ integration: Semantic models are now the core interface for agent reasoning
  • Optimization for Copilot: Well-documented models with clear descriptions dramatically improve AI accuracy

New Power Query Experience (Preview)

Enhanced Power Query interface in Power BI Desktop for more intuitive data shaping, transformation, and preparation workflows.

  • Modernized UI for faster data transformation
  • Improved discoverability of functions and connectors
  • Better handling of complex transformations and error diagnostics

Report & Formatting Updates (April–May 2026)

Recent monthly releases added fine-grained controls and quality-of-life improvements for report authors.

  • Exact pixel sizing for card visuals
  • Advanced theme switching and new visual types
  • AI-powered summarization shortcuts in report headers
  • Expanded AI-driven features across Desktop and Service
  • Deprecation of some legacy capabilities — modernize to Copilot-native workflows
💡 Best Practice: Optimize Your Semantic Model for the Agentic Era

With agents and Copilot now central to Power BI, invest in model quality: add clear descriptions to tables/columns, define DAX UDFs for common calculations, use consistent naming conventions, and configure AI Instructions. Clean, well-documented models yield dramatically better AI results.

📚 Learn More

Build 2026: Fabric & Microsoft Databases Blog ↗ Building the Agentic Analytics Stack ↗ Rayfin SDK (GitHub) ↗ Power BI Updates Blog ↗ Fabric IQ Overview ↗ Build 2026 News (GitHub) ↗ Calling DAX UDFs from Copilot (Chris Webb) ↗
FabCon 2026

FabCon & SQLCon 2026 News

Key announcements from the third annual FabCon and first-ever SQLCon in Atlanta — unifying databases and Fabric on a single data platform.

📅 FabCon & SQLCon 2026 — Atlanta, Georgia

Nearly 300 workshops and sessions across 8,000 attendees. Microsoft Fabric now serves 31,000+ customers — the fastest-growing data platform in Microsoft's history. SQL Server 2025 is growing more than 2× faster than the previous version.

Major Announcements

Database Hub in Fabric

A unified database management experience bringing Azure SQL, Cosmos DB, PostgreSQL, MySQL, SQL Server (via Arc), and Fabric Databases into a single pane. Features agent-assisted, human-in-the-loop management with Copilot-powered insights.

  • Unified observability & delegated governance
  • Intelligent agents to surface what changed and guide action
  • New savings plan for databases — up to 35% savings

OneLake Expansion

Mirroring now supports SharePoint Lists (preview), Dremio (preview), Oracle and SAP Datasphere (GA), and Azure Monitor (preview). Extended capabilities include Change Data Feed and views on mirrored data.

  • Shortcut transformations — GA (Excel → Delta tables in preview)
  • OneLake security — GA (row/column-level controls)
  • Read from OneLake via Azure Databricks Unity Catalog (preview)
  • Snowflake interoperability — GA

Fabric Analytics & Runtime 2.0

Runtime 2.0 (preview) brings Apache Spark 4.x, Delta Lake 4.x, and Scala 2.13 for large-scale data computation. Materialized lake views are now GA for simplified medallion architecture implementation.

  • Materialized Lake Views (GA) — precomputed, auto-refreshed views over lakehouse data. Define a SELECT statement that spans Bronze/Silver/Gold and Fabric keeps the result materialized and fresh — replacing many manual notebook pipelines
  • Graph in Fabric (GA) — native graph database with GQL (Graph Query Language) support for modeling entity relationships like supply chains, customer networks, and knowledge graphs at scale
  • Agentic Copilot in notebooks — workspace-aware code generation
  • Maps in Fabric — GA for geospatial real-time insights

Fabric IQ & Semantic Intelligence

Fabric IQ ontologies accessible via MCP server (GA), enabling agents to discover and act on your semantic layer.

  • Planning in Fabric IQ — new enterprise planning capability to create plans, budgets, forecasts, and scenario models directly on semantic models. Combines historical, real-time, and forward-looking data in a single source.
  • Direct Lake on OneLake — GA (import-class performance, no data movement)
  • Graph in Fabric — GA (visualize complex entity relationships)
  • NVIDIA partnership for Physical AI & digital twins

Data & Operations Agents

Fabric data agents are now GA — virtual analysts aligned to specific domain data for deeper analysis and insights via natural language. Operations agents monitor real-time data, detect patterns, and take proactive action.

  • Usable across Fabric, Microsoft Foundry, Copilot Studio, and M365 Copilot
  • Grounded in ontology for business-context-aware responses

Power BI Translytical Task Flows

Now GA — users can trigger operational workflows, resolve data issues, and take action directly from Power BI reports. Elevates Power BI from analysis-only to an action-oriented platform.

  • Embed task flows in dashboards — click a KPI to kick off a business process
  • Connects to Power Automate, Dataverse, and custom APIs
  • Audit trail for all actions taken from reports

Fabric Planner (GA)

Enterprise planning and budgeting built directly into Fabric. Create plans, budgets, forecasts, and scenario models on top of your semantic models — combining historical, real-time, and forward-looking data in one governed environment.

  • No-code/low-code "what-if" scenario analysis with ML-powered forecasting
  • Replaces spreadsheet-based planning with governed, auditable workflows
  • Integrated into Fabric IQ semantic layer for business-context-aware planning

Developer Experience

Fabric local MCP is now GA — connecting AI coding assistants like GitHub Copilot directly to Fabric. Fabric remote MCP (preview) enables cloud-hosted agent execution with authenticated actions.

  • Selective Branching (preview) — branch only specific items instead of full workspaces, reducing clutter and improving dev velocity
  • Variable Library (preview) — centralized configuration management for deployment pipelines, replacing hard-coded environment references across Dev/Test/Prod
  • Git selective branching for feature-scoped workflows
  • Agent Skills for Fabric — open-source GitHub Copilot terminal plugins
  • Fabric Jumpstart — reference architectures & one-click sample deployments
  • Fabric Extensibility Toolkit (FET) — GA with CI/CD support
  • Fabric CLI v1.5 — one-command workspace deployments with AI and Power BI integration

Migration Assistants

New migration assistants (public preview) for Azure Data Factory, Synapse Analytics, and Azure SQL help move pipelines, Spark pools, notebooks, and database schemas into Fabric with minimal disruption.

  • ADF & Synapse: evaluate, convert, and optimize pipelines incrementally
  • SQL: import schemas via DACPACs with AI-assisted compatibility resolution

AI-Ready Data Foundation — 4 Core Steps

💡 Building a Complete AI Foundation

Microsoft IQ provides an intelligence layer delivering shared, enterprise-grade business context from three sources: Work IQ (productivity signals), Foundry IQ (institutional knowledge), and Fabric IQ (live business data).

1. Unify Your Data Estate

Eliminate silos and reduce architectural complexity with OneLake as a single logical data lake — no unnecessary ETL, fragmentation, or duplicated copies.

2. Process & Harmonize Data

Make data AI-ready — clean, connected, and structured for both operational and analytical use with Spark, T-SQL, KQL, and Analysis Services engines.

3. Curate Semantic Meaning

Give agents contextual understanding through Fabric IQ ontologies, enabling them to interpret data the way your teams already do.

4. Empower AI Agents to Act

Apply business context to automate workflows, accelerate decisions, and transform operations with data agents and operations agents.

📚 Learn More

FabCon & SQLCon 2026 Keynote Blog ↗ Database Hub in Fabric ↗ Fabric IQ Overview ↗ Fabric Data Agents ↗ Fabric MCP Server (GitHub) ↗ Fabric Jumpstart ↗
Updates

What's New

Latest Microsoft Fabric blog posts, automatically categorized by workload — powered by the official Fabric blog RSS feed.