Microsoft and Databricks are set to use the Data + AI Summit 2026, taking place June 15–18 in San Francisco and virtually, as a launchpad to reinforce Azure Databricks as the governed enterprise platform for data engineering, analytics, and AI agent development. The conference arrives at a moment when organizations are racing to deploy AI agents that interact with sensitive corporate data, making governance not just a feature but a business requirement. This year’s summit will showcase how Azure Databricks combines the Unity Catalog, OneLake integration, and Microsoft Fabric to create a unified, governed environment for building and managing AI agents at scale.

Why Governance Matters for AI Agents

AI agents are autonomous software entities that reason, plan, and act on behalf of users. They pull data from lakehouses, apply models, and trigger actions across enterprise systems. Without rigorous governance, an agent can inadvertently expose personally identifiable information, violate compliance mandates, or make decisions based on stale, unauthorized data sources. A recent industry survey found that 68% of data leaders cite governance as the primary barrier to deploying production AI agents. Microsoft and Databricks aim to remove that barrier by embedding governance directly into the agent development lifecycle.

Azure Databricks already serves thousands of enterprises that run massive ETL pipelines, machine learning workloads, and real-time analytics. The platform’s shift toward agent governance responds to a clear market need: teams need a single control plane to manage data access, model lineage, and agent behavior. At the summit, attendees will see how Unity Catalog extends its reach beyond tables and files to encompass AI models, functions, and agent metadata.

Unity Catalog: The Governance Backbone

Unity Catalog provides a centralized governance solution for data and AI assets across clouds. It captures fine-grained lineage for every asset—from raw data ingestion through feature engineering, model training, and agent execution. Administrators can set row-level and column-level access controls, tag data with sensitivity labels, and audit all access attempts. For agent development, this means data scientists can define an agent’s allowed data surface declaratively. If a regulatory obligation requires masking credit card numbers, Unity Catalog enforces that mask at query time, regardless of whether the caller is a human analyst or an AI agent.

At the summit, Microsoft and Databricks will demonstrate how Unity Catalog’s model registry integrates with agent frameworks. Teams can register a model version, attach compliance notes, and instantly propagate those policies to any agent that invokes the model. This closed-loop governance ensures that an agent cannot call a deprecated model or one that has not passed a fairness review. In addition, the catalog tracks agent-specific metrics, such as response accuracy and data freshness, giving governance officers a real-time dashboard of agent health.

OneLake Integration and Microsoft Fabric

OneLake, the single data lake that underpins Microsoft Fabric, becomes a native home for Azure Databricks data. The integration means that tables, views, and shortcuts from OneLake appear automatically in Unity Catalog, eliminating redundant copies and sync delays. An agent running in Azure Databricks can query Fabric data through OneLake shortcuts with the same access policies applied uniformly. This alignment between Databricks and Fabric simplifies cross-team collaboration: data engineers working in Fabric notebooks and data scientists building agents in Databricks share one governed copy of the data.

At Data + AI Summit 2026, expect deep-dive sessions on how OneLake shortcuts accelerate agent development. A common pattern is for an organization to keep its gold-level curated datasets in OneLake, enrich them with Azure Databricks’ Spark compute, and then serve features to agents through Unity Catalog. Governance policies flow automatically from Fabric’s sensitivity labels to Databricks, ensuring that any agent accessing a shortcut inherits the correct classification and handling rules. Microsoft has confirmed that this integration will be demonstrated live, showing an end-to-end scenario of an agent responding to a natural language query about sales performance without ever bypassing data masking rules.

Building Governed Agents in Azure Databricks

The summit will offer hands-on labs that guide participants through building a governed AI agent on Azure Databricks. The workflow begins with discovering trusted datasets via Unity Catalog’s search and lineage graph. A developer selects a table, reviews its quality score and owner, and registers it as an approved data source for an agent. Next, the developer authors a Python notebook that uses a large language model to interpret user queries and generate SQL against that table. Unity Catalog’s row-level security applies transparently, so the agent can only see data within the user’s scope.

To prevent prompt injection or indirect data leakage, Databricks has added guardrails that inspect agent-generated code before execution. The system compares the planned query against the agent’s registered data surface and rejects any attempt to access unapproved tables or columns. All audit logs feed into Microsoft Purview, allowing compliance teams to monitor every interaction. The summit demos will highlight this integration, showing how an agent’s attempt to query a masked column triggers a real-time alert and blocks execution.

Expanding Governance to Multi-Agent Workflows

Single agents are giving way to multi-agent systems that collaborate to solve complex tasks. Azure Databricks’ governance model scales to these scenarios by treating each agent as a principal with its own identity and set of entitlements. Unity Catalog can secure inter-agent communication, ensuring that an agent orchestrating a workflow can only pass data to a downstream agent that is authorized to receive it. This fine-grained control prevents privilege escalation and data spillage between agents.

During the summit, a dedicated session titled “Governance for Multi-Agent AI” will explore how enterprises can model agent identities, share data objects, and monitor cross-agent data flows. Using a supply-chain example, presenters will show how an inventory-forecasting agent can securely exchange data with a procurement agent while a governance steward reviews the shared data lineage in real time. This represents a significant leap from traditional data governance, which was designed for human users and static dashboards.

Real-World Impact on Enterprise Workloads

Microsoft and Databricks intend to prove that governed agent development is not just theoretical. Several customer stories will be shared, detailing how Azure Databricks’ governance features helped meet strict regulatory requirements. One financial services firm used Unity Catalog to build a mortgage advisory agent that explains denial reasons while redacting sensitive financial details. A healthcare provider developed a patient-triage agent that respects HIPAA data-segmentation rules by leveraging column-level masks and user-attribute-based access. These use cases underscore that governance is not a bottleneck but an enabler for agent adoption.

The summit will also address performance considerations. A governed agent must not add latency that ruins the user experience. Databricks’ serverless compute and improved query compilation ensure that enforcement checks complete in sub-second time, even on petabyte-scale tables. Presenters will share benchmarks showing that row-level security on OneLake data adds less than 5% overhead compared to an unsecured query, making governance practical for interactive agent use.

What This Means for Windows and Microsoft Enthusiasts

For Windows-focused professionals, Azure Databricks governance ties directly into the broader Microsoft ecosystem. The integration with Microsoft Purview, Entra ID, and Fabric means that tools many Windows admins already manage become the foundation for AI agent security. Group policies, conditional access, and authentication flows extend naturally to Databricks agents. This alignment reduces the learning curve and allows organizations to leverage existing investments in Windows Server, SQL Server, and Power Platform when building modern AI solutions.

Windows developers building agent-enabled applications will benefit from the open-source frameworks that Databricks supports, such as LangChain and Semantic Kernel. These frameworks can register agents in Unity Catalog, enabling enterprise-grade logging and monitoring with minimal code changes. Summit sessions will include code examples in C# and Python, demonstrating that governance is accessible to Windows dev teams as well as data engineers.

Looking Ahead: The Roadmap for Azure Databricks Governance

Beyond the summit, the joint roadmap includes tighter integration with Microsoft Copilot and Azure AI Studio. Plans are underway to let Copilot answer governance questions directly—such as “Who accessed this dataset through an agent last week?”—by querying Unity Catalog’s audit logs. Meanwhile, Azure AI Studio users will be able to publish their agents to the Databricks Unity Catalog, making them discoverable and governed from one pane. This cross-product synergy signals that Microsoft is betting big on a unified data + AI governance plane.

The Data + AI Summit 2026 will serve as the forum where these capabilities move from slideware to live demos. Attendees can register for hands-on workshops that walk through setting up governed agents on real clusters. Virtual participants will have access to on-demand sessions covering best practices, security patterns, and performance tuning.

For any organization still hesitant to deploy AI agents due to governance fears, Microsoft and Databricks are delivering a clear answer: the technology is ready, it integrates with existing compliance frameworks, and it can be adopted incrementally. The coming announcements will reinforce that Azure Databricks is not just a data platform but the control center for an enterprise’s entire AI agent fleet.