Enterprises have spent the past two years running AI pilots. Most produced results in sandboxes that never reached production. The gap was rarely the model. It was fragmented data, undefined governance, and infrastructure never built for agents at scale. At Databricks Data + AI Summit 2026, held in San Francisco, Databricks addressed all three with a wave of releases, from Genie Ontology and Unity AI Gateway to OpenSharing and a rebuilt Agent Bricks platform.
In this article, we’ll cover the eight most significant announcements from DAIS 2026, what each one means for enterprise teams, and how Kanerika helps you act on them.
Key Takeaways LTAP unifies transactional and analytical data on a single open-format storage layer, removing ETL pipelines for good. Lakehouse//RT delivers sub-100ms query latency directly on Delta and Iceberg tables , up to 16x faster than a separate real-time stack. Agent Bricks is now a full infrastructure platform: 100,000+ agents built, 1 quadrillion+ tokens processed annually. Genie One (GA) uses Genie Ontology to give agents real business context: what “revenue” or “churn” actually means in your org. Unity AI Gateway governs what agents do at runtime, with spend caps, PII guardrails, and a security partner ecosystem. OneLake Catalog Federation is now GA, making a unified Databricks and Microsoft Fabric data estate a production reality.
Top 8 Announcements from Databricks Data + AI Summit 2026 1. LTAP: Unifying Transactional and Analytical Data For four decades, enterprise data meant two separate systems: a transactional database for applications and an analytical platform for reporting. Syncing the two required ETL pipelines , CDC connectors, and replication jobs that added cost and latency at every step.
Databricks’ LTAP (Lake Transactional/Analytical Processing) closes that gap. Operational data written to Lakebase (Databricks’ serverless Postgres) is stored directly in Unity Catalog using Delta and Iceberg formats from the point of write. No sync, no hidden copy, no format conversion.
Enterprise impact: AI agents no longer have to wait for pipeline schedules to act on current data. Every engine reads the same governed copy. This is the foundation that makes real-time, trusted AI action on enterprise data possible.
Alongside LTAP, Databricks expanded Lakebase (now at 12 million database launches per day) with:
Cross-cloud and cross-region disaster recovery (GA) Git-style branching and snapshots for safe experimentation (GA) Autonomous database operations for agent-assisted health monitoring (GA) Lakebase Search: hybrid vector and full-text retrieval built into Postgres (Beta) Status: LTAP is coming soon as part of Lakebase. Enterprise Lakebase upgrades are GA now.
LTAP and Lakebase Capabilities at DAIS 2026 LTAP Property What It Delivers Status Unified storage (Delta/Iceberg) One governed copy for transactional, analytical, and streaming workloads Coming soon Lakebase DR (cross-cloud/region) Resilient architecture for mission-critical agent operations GA Git-style branching and snapshots Safe experimentation on production data without risk GA Autonomous database operations Agents monitor health, propose indexes, assist recovery GA Lakebase Search Hybrid vector and full-text retrieval built into Postgres Beta
2. Lakehouse//RT: Real-Time Queries Directly on the Lakehouse (Beta) Running real-time analytics has traditionally meant maintaining a separate serving tier alongside the lakehouse. More cost, another governance boundary, and duplicated data.
Lakehouse//RT , powered by a new engine called Reyden, removes that requirement. It delivers sub-100ms query latency at up to 12,000 queries per second directly on governed Delta Lake and Iceberg tables, up to 16 times faster than a separate real-time stack.
Enterprise impact: Teams running operational dashboards, fraud detection, or real-time agent workflows no longer need a dedicated serving layer. The data stays in the lakehouse. Governance stays intact.
Mehrshad Setayesh, SVP Engineering at PointClickCare, noted Lakehouse//RT ran more than a third faster than their prior warehouse on their healthcare dataset and removed the need for a dedicated real-time system entirely.
3. Agent Bricks: From Toolkit to Full Agent Infrastructure Platform (GA) Agent Bricks launched at DAIS 2025. In the year since, more than 100,000 agents have been built on it, processing over 1 quadrillion tokens annually. AstraZeneca, 7-Eleven, Fox Corporation, and Block all run agents on Agent Bricks in production.
At DAIS 2026, Databricks expanded it significantly. The core realization: building the agent loop is 1% of the work. The other 99% is infrastructure: memory, deployment, security, evaluation, monitoring, and context. Agent Bricks now covers all of it.
What’s new:
Managed agent memory (Lakebase-powered): Agents persist context across sessions and share it across instancesMCP-connected retrieval: Agents connect to Unity Catalog data plus external tools like Google Drive, Jira, Slack, and GitHubDocument Intelligence: PDF and document parsing, now GASecure sandboxes: Isolated compute environments for agent tool useMulti-model choice: OpenAI, Anthropic, Gemini, Qwen, Kimi, and Grok, all governed under Unity CatalogLakeWatch integration: Agent traces and outputs land in the lakehouse for security monitoring and PII alertsEnterprise impact: Teams get an end-to-end platform rather than assembling their own agent infrastructure from scratch. Read our detailed breakdown: Databricks Agent Bricks explained .
Agentic AI Development and Deployment Kanerika designs and deploys production-grade AI agents across analytics, document intelligence, compliance, privacy, validation, and voice domains, with governance architecture built in from the start.
Explore Agentic AI Services →
4. Genie One and Genie Ontology: Business AI That Actually Understands Your Data (GA) Most enterprise AI tools can summarize documents or answer generic questions. What they cannot do is explain why margins dropped last quarter or which accounts carry the highest renewal risk. That context is scattered across dozens of systems.
Genie One is Databricks’ answer. Now generally available, it connects to 50+ applications including Google Drive, Jira, Teams, Confluence, and SharePoint. It can answer questions, run analysis, monitor changes, and trigger workflows across both structured lakehouse data and unstructured content.
The intelligence layer underneath is Genie Ontology , a continuously learning context layer that reads your data, documents, and applications over time to understand what your business actually means by “revenue,” “churn,” or “active user.” That is what separates a plausible answer from a correct one.
Enterprise impact: Business teams get an AI coworker grounded in real company data, not a generic chatbot. No seat-based pricing: up to $10 per user per month in AI credits.
Read more: Databricks Genie explained .
5. Genie Code: Autonomous Data and ML Engineering (GA) Genie Code is the agentic coding assistant embedded in Databricks notebooks and pipelines. DAIS 2026 brought three key upgrades:
Full-page command center: Manage multiple parallel threads across data engineering and ML workflows without losing contextFull ML stack integration: Reads experiment data from MLflow, inspects model serving endpoints, handles compute setup automaticallyScheduled tasks (coming soon): Genie Code runs analysis on a schedule and delivers results for human review, moving from interactive to autonomousRadu Dragusin, Principal Engineer at Danfoss, described going from raw data to a production-ready ML workflow in 90 minutes: training models, registering them with MLflow and Unity Catalog, and deploying to a serving endpoint within a single session.
6. Unity AI Gateway: Runtime Governance for Enterprise Agents (GA) Traditional AI governance focused on data access: who can see which table, which model is approved. That works for human users. It breaks down for agents that act autonomously, call tools, spawn subagents, and generate outputs at volume across production systems.
Unity AI Gateway governs what agents do at runtime, not just what they can access. It is built on Unity Catalog and covers:
Hard spend caps and smart routing to control AI costs across providersReal-time guardrails for PII detection and prompt injection preventionFull tracing of agent actions, tool calls, and MCP interactionsStateful security policies written in SQL (soon Python). These policies respond to what an agent is doing in context, not just applying a static ruleDatabricks has opened Unity AI Gateway to a runtime security ecosystem including CrowdStrike, Cyera, HiddenLayer, Netskope, and Noma Security.
Enterprise impact: Regulated industries finally have the governance architecture to run agents in production, not only in pilots. Kanerika’s AI governance practice is built around Unity Catalog and Microsoft Purview as the governance foundations for exactly this kind of deployment.
Case Study: Real-Time Compliance and Risk Detection Through an AI Agent See how Kanerika built a real-time compliance AI agent for a regulated client, with continuous risk monitoring , automated flagging, and audit-ready decision trails built into the architecture
Read the Case Study
.7. OpenSharing: Open Protocol for Data, AI, and Agent Sharing (GA) Delta Sharing proved enterprises would choose open data sharing over proprietary lock-in. OpenSharing extends that to the full AI stack: data, models, agent skills, and governed AI experiences can now be shared across any format, any cloud, and any organizational boundary.
OpenSharing is now an open-source protocol hosted by the Linux Foundation, supporting Delta Lake, Apache Iceberg, and Parquet. Matei Zaharia, Co-founder and CTO of Databricks, described it as extending the open-over-locked-in principle to the entire AI stack.
Key capabilities for enterprise teams:
Genie Agent Sharing: Share governed AI experiences, including semantic context, business metrics, and reusable logic, with external partners, with controls on access, quotas, and proprietary instruction visibilitySecureConnect: A Databricks-managed proxy that routes storage access for all recipients without per-recipient firewall changes, removing the manual networking overhead that made large-scale sharing impracticalEnterprise impact: Organizations can now monetize data and AI assets on a usage-based model, share governed agents with partners, and handle multi-cloud data exchange without the infrastructure overhead.
8. Azure Databricks and Microsoft Fabric: The Unified Estate Is Now GA Organizations running both Azure Databricks and Microsoft Fabric have had to choose between two data estates or build expensive pipelines between them. At DAIS 2026, that changes.
OneLake Catalog Federation is now generally available. Azure Databricks can query data stored in OneLake directly through Unity Catalog, with no data movement required. This is the foundational piece that makes a true unified estate possible.
Additional features at various stages:
Managed Delta tables in OneLake (Beta): Data written by Databricks is natively available to all Fabric engines, zero-copyGenie in Microsoft Teams and M365 Copilot (Beta): Lakehouse-grounded AI inside the collaboration tools where enterprise teams already workAzure Databricks Excel Add-in (Public Preview): Query lakehouse data directly in ExcelSharePoint Connector (Beta): Automate ingestion of SharePoint files into the lakehouseEnterprise impact: Teams running data workloads across both platforms can now treat them as a single governed estate rather than managing two separate environments. As both a certified Databricks partner and a Microsoft Solutions Partner for Data and AI , Kanerika is one of the few partners built to execute on this convergence.
Azure Databricks and Microsoft Fabric Integration Status, June 2026 Feature Status at DAIS 2026 What It Enables OneLake Catalog Federation GA Databricks queries OneLake data through Unity Catalog, zero-copy Managed Delta tables in OneLake Beta Databricks-written data available to all Fabric engines, no movement Genie in Teams and M365 Copilot Beta Lakehouse-grounded AI inside daily enterprise collaboration tools Azure Databricks Excel Add-in Public Preview Query lakehouse data in Excel without leaving the spreadsheet SharePoint Connector Beta Automated ingestion of SharePoint files into the lakehouse
Other Notable Platform Security and Identity Updates As AI tools reach more enterprise users, identity management becomes a bottleneck. Databricks addressed several gaps at DAIS 2026:
Automatic Identity Management (AIM) for Entra ID: Now GA on AWS and GCP, automatically provisioning users, groups, and service principals from your identity provider, removing manual SCIM syncAIM for Okta: Now in public preview on AWS and GCPContext-Based Ingress (CBI): Public preview on all three clouds. Enables zero-trust access policies based on network, identity, and scope, so Genie and dashboards can be exposed to external users without opening the full workspacePrivate Network Gateway: A single, managed connection for serverless workloads to reach private data sources and internal APIs
How Kanerika Helps You Execute on Databricks Kanerika is a certified Databricks consulting partner and a Microsoft Solutions Partner for Data and AI with advanced specializations in Analytics on Azure and Data Warehouse Modernization. With 10+ years of delivery experience and a 98% client retention rate across 100+ enterprise clients, the team brings real implementation depth, not surface-level platform familiarity.
What Kanerika Brings to Every Engagement Kanerika operates across the full Databricks stack with capabilities that map directly to what DAIS 2026 announced:
Data engineering and migration: The FLIP platform automates migrations from Informatica, Tableau, Cognos, QlikView, SSIS, Crystal Reports, and others to Databricks, Microsoft Fabric, Power BI, and Snowflake, delivering 50 to 60% reduction in migration effort and 75% reduction in annual licensing costs on confirmed deployments. Available on the Azure Marketplace.Agentic AI delivery: Production AI agents deployed in live enterprise environments across financial services, healthcare, manufacturing, logistics, and retail, across analytics, document intelligence, compliance, privacy, validation, and voice domains.AI governance architecture:Databricks and Fabric unified practice: One of the few partners who can execute across both platforms simultaneously, which is exactly what the OneLake Catalog Federation GA now demands.ISO 27001, SOC 2 Type II, and CMMI Level 3 certifications form the compliance baseline every regulated client requires before production deployment.
Case Study: Context-Aware AI Agent for Expert Recommendations See how Kanerika built a governed, source-grounded AI agent for a financial services client, with zero hallucination incidents post-deployment and a traceable audit trail for every recommendation.
Read the Case Study
Dual Funding Advantage: Microsoft and Databricks Together Most implementation partners can help a client access funding from one platform. Kanerika’s dual standing means both funding levers can be applied to the same engagement.
Azure ECIF and MACC co-fund discovery, proof of value, and migration work on the Microsoft sideDatabricks DCIF and Velocity incentives fund proof of value, consumption ramp, and growth on the Databricks sideFor an enterprise modernization project touching both platforms, this lowers the total cost-to-modernize in a way single-alliance partners structurally cannot match. It also creates a faster consumption path that activates Velocity rebates over time.
Ready to Build on Databricks? Whether you are planning a migration to Databricks, designing a governed agentic AI architecture , or evaluating the Fabric-Databricks unified estate, our certified team can help you.
Book a Meeting
Wrapping Up DAIS 2026 was a consolidation event. Databricks tied data infrastructure, AI development, business AI, governance, and open sharing into one platform aimed at a single outcome: agents that work reliably in enterprise production. Every announcement addressed the same constraint. Agents are only as good as the context, governance, and architecture underneath them. The GA releases at this summit built concrete answers to all three. For enterprise teams ready to move from pilot to production, the infrastructure case is now substantially stronger. The next step is execution, and that is where implementation expertise matters most.
FAQs What was announced at Databricks Data + AI Summit 2026? Databricks announced eight major updates at DAIS 2026, held June 15 to 18 in San Francisco: LTAP (unifying transactional and analytical data on open formats), Lakehouse//RT (real-time query engine), expanded Agent Bricks platform (now GA), Genie One (agentic business AI, GA), Unity AI Gateway (runtime governance, GA), OpenSharing (open data and AI sharing protocol), OneLake Catalog Federation GA for the Azure Databricks and Microsoft Fabric convergence, and platform security updates including Automatic Identity Management on AWS and GCP.
What is LTAP in Databricks? LTAP (Lake Transactional/Analytical Processing) is a new data architecture from Databricks that unifies transactional and analytical workloads on a single open-format storage layer (Delta and Iceberg) under Unity Catalog governance. It eliminates the ETL pipelines that have historically separated operational databases from analytical platforms, giving AI agents a single, current, governed data surface to act on. LTAP is built on Lakebase and is listed as coming soon.
What is Genie One in Databricks? Genie One is Databricks’ agentic coworker for business teams, now generally available. It connects to 50+ applications including Google Drive, Jira, Slack, Teams, and Confluence, and can answer questions, run analysis, monitor changes, and trigger workflows across both structured lakehouse data and unstructured content. Its context layer, Genie Ontology, continuously learns the business meaning behind the organization’s data, making answers grounded rather than generic.
What is Unity AI Gateway? Unity AI Gateway is Databricks’ runtime governance layer for enterprise AI, generally available as of DAIS 2026. Built on Unity Catalog, it governs what agents actually do at runtime, including tool calls, MCP interactions, and outputs, with real-time guardrails, hard spend caps, smart routing across providers, and full tracing. Security policies are written in SQL and can respond dynamically to agent behavior in context. Databricks has opened it to a partner ecosystem including CrowdStrike, Cyera, and HiddenLayer.
Is Azure Databricks integration with Microsoft Fabric ready for production? OneLake Catalog Federation, which lets Azure Databricks query OneLake data through Unity Catalog without copying data, is now generally available. Managed Delta tables in OneLake are in public beta. These two features together make the unified Fabric-Databricks data estate a production starting point rather than a roadmap item, and they are directly relevant to organizations already running data workloads across both platforms.
What is Agent Bricks in Databricks? Agent Bricks is Databricks’ full-stack platform for building, deploying, and governing enterprise AI agents, now generally available. It covers the full 99% of agentic infrastructure: managed memory via Lakebase, MCP-connected context retrieval from Unity Catalog and external sources, multi-model support, secure sandboxes, LakeWatch security integration, and built-in evaluation and monitoring. More than 100,000 agents have been built on the platform, with over 1 quadrillion tokens processed annually.
What is OpenSharing in Databricks? OpenSharing is an open protocol, now hosted by the Linux Foundation, for sharing data, AI agents, models, and skills across organizations and clouds. It supports Delta Lake, Iceberg, and Parquet natively. The Databricks enterprise implementation adds Unity Catalog governance and Genie Agent Sharing, which lets organizations share governed AI experiences (including business context and reusable logic) with partners, with controls on access, quotas, and data visibility. SecureConnect removes the per-recipient networking overhead that made large-scale sharing impractical.
How can Kanerika help enterprises act on DAIS 2026 announcements? Kanerika is a certified Databricks consulting partner and Microsoft Solutions Partner for Data and AI with active specializations in Analytics on Azure and Data Warehouse Modernization. The team delivers Databricks migrations using the FLIP accelerator platform, designs governed agentic AI architectures using Unity Catalog and Unity AI Gateway, and operates as a dual Databricks and Microsoft Fabric practice, relevant for the OneLake federation convergence most large enterprise data teams will need to plan for.
