Overview
Azure Databricks announced advances across four pillars at DAIS 2026: a new data architecture for the agentic era, Genie expansion across the Microsoft ecosystem, a lakehouse-native customer data platform, and a governance framework covering everything from semantic context to runtime cost control.
Deep integration with the Microsoft ecosystem — M365 Copilot, Teams, Excel, SharePoint, OneLake — is the central differentiator, positioning Azure Databricks as the unified data and AI layer for organizations already living in the Microsoft world.
Agentic Data: LTAP Architecture
Azure Databricks introduced the industry’s first LTAP (Lake Transactional/Analytical Processing) architecture, unifying analytical data, streaming pipelines, and live application transactions in a single storage layer without data replication.
Lakebase: Serverless Postgres database with instant copy-on-write branching for safe production debugging. Lakehouse//RT delivers sub-second, millisecond-level response times for high-concurrency workloads using the Reyden engine. OneLake Integration: Zero-copy data access across the Microsoft Fabric ecosystem (GA for querying, Public Beta for storing).
PointClickCare, a healthcare customer, reported their queries run “more than a third faster on average” with “10× faster queries” compared to previous warehouse solutions.
Genie Across the Microsoft Ecosystem
Databricks expanded its AI assistant across Microsoft’s ecosystem to make intelligence accessible within daily collaboration tools:
Genie for Teams & M365 Copilot (Beta): Direct data queries within Teams conversations and M365 workflows. Excel Add-in (Public Preview): Native lakehouse access without SQL or per-user setup; supports Unity Catalog metric views. SharePoint Connector (Public Beta): Automated ingestion of structured and unstructured files (PDFs, Word, PowerPoint).
The complete Genie framework: Genie One (AI coworker for business teams), Genie Agents (contextual conversation agents for non-technical users), Genie App Builder (low-code environment for custom applications), Lakeflow Designer (natural-language pipeline orchestration), Genie ZeroOps (autonomous infrastructure provisioning), and Genie Code (AI development partner).
userLake: Lakehouse-Native CDP
userLake represents the first lakehouse-native Customer Data Platform, eliminating fragmented MarTech complexity without duplicating or moving data outside the lakehouse.
Features include: Autonomous Profile Agents transforming fragmented data into unified Customer 360 profiles, Campaign Agents enabling audience segmentation and 1:1 personalization, autonomous next-best-action recommendations, and cross-channel activation capabilities.
Governance Framework: Context and Control
Genie Ontology: Self-improving semantic context engine automatically extracting table relationships, column metrics, and query signals. Eliminates manual curation and reduces AI hallucinations by capturing enterprise-specific terminology.
Unity AI Gateway: Centralized runtime registry within Unity Catalog. Real-time rate limiting and content filtering. Hard spend caps ensuring predictable token economics. Governs all automated workflows at runtime.
Key Points
- First industry LTAP architecture: analytical + transactional + streaming in one layer
- Lakebase: serverless Postgres database with instant branching for safe debugging
- Lakehouse//RT: sub-second responses for high-concurrency workloads with Reyden engine
- OneLake GA: zero-copy access to the Microsoft Fabric ecosystem
- Genie in Teams, M365 Copilot, and Excel Add-in for analytics in the workflow
- SharePoint Connector for automatic enterprise document ingestion
- userLake: first lakehouse-native CDP without moving data
- Profile Agents for automatic Customer 360 from fragmented data
- Genie Ontology eliminates manual curation and reduces hallucinations
Why It Matters
For organizations already using Azure and the Microsoft ecosystem (Teams, SharePoint, Excel, M365 Copilot), Azure Databricks closes a gap that had been frustrating: important information lives in the lakehouse, but day-to-day work happens in Teams and Excel. Moving information between those worlds was costly and error-prone.
Embedding Genie directly in Teams and Excel means a sales director can ask questions about team performance in the same channel where they already work — without learning a new tool, without waiting for a report from the data team, with the guarantee that the numbers come from the authoritative system.
userLake addresses a specific pain point of digital marketing: the proliferation of Customer Data Platforms (CDPs, DMPs, personalization tools) each maintaining their own copy of customer data. By collapsing these platforms within the lakehouse, it eliminates cross-system synchronization and ensures all marketing decisions are made on the same data that governs the rest of the business.
