Bridging Data Science and Marketing: Databricks Unveils Delta Sharing Integration for Adobe Experience Platform and Agentic Marketing Workflows

In today’s hyper-competitive landscape, "speed to insight" is no longer the finish line. The new gold standard is speed to activation.

For years, a massive gap has separated where customer intelligence is managed and where marketing is executed. Data teams build sophisticated models and metrics in the Lakehouse, from propensity scoring to Customer Lifetime Value (CLV). Marketing teams execute campaigns and journeys in purpose-built applications such as Journey Optimizer and Real-Time CDP, built on Adobe Experience Platform. 

Until recently, bridging that gap has meant pipelines, duplication, and delay. Marketers have been stuck in a cycle of waiting for CSV exports, scheduling SFTP transfers, and maintaining expensive ETL pipelines just to activate their own customer data in Experience Platform.

Today, Databricks and Adobe are closing the gap. We’re announcing three major milestones in our partnership:

1. Delta Sharing support for Adobe Experience Platform, enabling direct, zero-copy access to governed first-party data in Databricks.
2. Databricks Genie connection with Adobe Marketing Agent via Model Context Protocol (MCP), making operational intelligence in Databricks available to Marketers without leaving the Experience Platform. 
3. An upcoming Adobe Marketing Agent beta release for the Databricks Marketplace, allowing Databricks customers to build supervisor agents that orchestrate across Genie and Adobe’s Marketing Agent to deliver connected insights for campaign performance, customer journey analysis, and business impact measurement.

Together, these releases establish a new foundation for marketers where data, activation, and AI operate on shared, real-time context without the overhead of moving or duplicating data.

This marks a significant step forward in the strategic partnership between Databricks and Adobe, accelerating our ability to deliver more connected, intelligent customer experiences.

The Challenge: Why this matters now

AI is accelerating how campaigns are created, optimized, and executed. At the same time, customer expectations for relevance and timing are higher than ever. But most marketing architectures weren’t built for the reality of AI agents acting across context and systems in real time.

The result:

• Insights arrive too late to inform campaign execution, leading to inefficient budget allocation and limiting the impact of predictive models and experimentation.
• Customer context is fragmented across systems, making timely, consistent personalization difficult and expensive.
• Data must be copied to be activated, increasing infrastructure cost and introducing governance and compliance risk.

Meanwhile, marketing teams are looking to improve personalization and targeting by leveraging the refined, governed, and valuable AI-enriched datasets already living in Databricks. At the same time, technical teams are under pressure to accelerate data access and time-to-insight while reducing duplication and infrastructure costs.

What’s missing is a way to activate data directly in real-time, without the cost, latency, and risk of moving data between systems. 

To address this, we’ve partnered closely with Adobe to bring two new major capabilities to market:

Zero-copy data access with Delta Sharing for Adobe Experience Platform

With Delta Sharing, Experience Platform can now directly access data in Databricks without ETL pipelines, duplication, or delay. This solves three of the biggest constraints in modern marketing:

• Latency: Insights in Databricks often take 24-48 hours to reach downstream marketing systems.
• Cost: Moving petabytes of data across clouds creates egress fees and redundant storage costs.
• Governance: Copying data into other platforms breaks lineage, access controls, and consistency defined in Unity Catalog. 

This partnership enables a Direct-to-platform sharing flow. Instead of "pushing" files from Databricks to Adobe, Experience Platform now acts as a native Delta Sharing recipient.

How it works

1. Unity Catalog as the source: Data teams curate governed tables and views in Databricks, such as high-value audiences, churn risk scores, or propensity models.
2. Secure Handshake: Databricks generates secure credentials using the open Delta Sharing protocol, allowing Experience Platform to directly access shared datasets.
3. Governed Direct Data Access: These datasets appear as virtual tables within Experience Platform, while Unity Catalog governs access. With Adobe Data Distiller, marketers can query live Databricks data in real time, without moving a single underlying record.

Built on open standards

This integration is powered by the open-source Delta Sharing protocol, not proprietary connectors or middleware. The result is a secure, governed, live connection between your Lakehouse and Adobe, with no data duplication required.

AI bidirectional orchestration with Adobe Experience Platform Agent Orchestrator and Genie MCP

While Delta Sharing solves the data access problem, the real challenge is making business context self-service. To address this organizational issue, we’re proud to announce an Adobe Experience Platform Agent Orchestrator integration with Databricks Genie MCP.

As part of this partnership, Databricks Genie MCP will be accessible from within the Adobe Experience Platform Agent Orchestrator, allowing Adobe users to interact with their relevant operational data in Databricks using natural language. And the Adobe Marketing Agent MCP will be accessible from within the Databricks platform, allowing for developers to deploy production-quality AI agents that incorporate AEP audience insights and engagement metrics to address cross-domain related use cases, such as closed-loop attribution.

What this unlocks

• Agentic intelligence across systems: Adobe’s AI agents can securely discover and use governed datasets, metadata, and models in Databricks via Genie’s MCP, to identify the best datasets for a specific campaign.
• Closed-loop execution back into Databricks: Databricks Genie and AI agents can query Adobe Marketing Agent using natural language for campaign performance, engagement, and conversion insights to retrain models, update segments, or trigger downstream workflows.
• Standardized, extensible integration: Using MCP, Adobe and Databricks agents can interact with Databricks tools like SQL Warehouses and Model Serving endpoints without custom integrations for each use case.

Examples

• Modeled data for audience creation and activation: A Databricks agent identifies a high-value micro-segment based on behavioral and product usage signals, then pushes that segment into Adobe to activate across paid and owned channels
• Continuous optimization loop: Campaign performance data from Adobe flows back into Databricks, where agents retrain models and improve future campaign performance without manual intervention

The Impact: Why this matters for joint customers

Eliminating the “integration data tax” allows organizations to reduce storage, egress, and engineering costs by activating data in place across environments like S3, ADLS, or GCS. 

At the same time, unified governance ensures consistent access controls and end-to-end lineage, so policies such as GDPR’s right to be forgotten are enforced seamlessly across both Databricks and Adobe Experience Platform. 

This foundation ultimately drives marketer empowerment, enabling teams to directly access and activate lakehouse data through agentic and Adobe Experience Platform interfaces without ongoing reliance on IT or data engineering.

A new foundation for modern marketing

Together, Databricks and Adobe are establishing a new foundation for how marketing systems work:

• Data stays centralized, governed, and trusted
• Activation happens without duplication
• AI operates on shared, real-time context

This is what enables marketing to move faster, operate more intelligently, and continuously improve.

Get started

If you’re attending Adobe Summit, visit us at Booth #548 to see the new integrations in action and learn how to connect your Databricks data to Adobe Experience Platform.

Sign up for the Delta Sharing & MCP waitlist to explore how you can start activating governed first-party data directly in Adobe. Join the waitlist here.