Eliminate the middleware tax: Real-time engagement with Braze and Databricks CustomerLake

TL;DR

Braze announced a broader bi-directional integration with Databricks to reduce the need for middleware tools and deliver a composable solution for customer engagement, bridging the gap between data and engagement.

Key takeaways

• Customers can use Braze and Databricks CustomerLake, an agentic CDP embedded natively within Databricks, to push their governed cohort audiences into Braze for segmentation
• Customers can also now use their governed, Databricks-hosted models within the BrazeAI Agent ConsoleTM to automate advanced conversational workflows and drive high-quality, hyper-personalized, 1-to-1 customer lifecycles at scale
• The Braze/Databricks integration includes support for Databricks Delta Sharing (beta), the most widely adopted open protocol for secure data sharing, to sync engagement data back to Databricks

Data and Marketing teams want the exact same thing: To give customers a flawless, deeply personalized experience. But behind the scenes, a quiet game of telephone has been undermining that experience for years.

Data teams build a pristine, comprehensive source of enterprise data inside data warehouses and lakehouses. Meanwhile, Marketing teams need to engage with customer behaviors the exact second they happen. To bridge the gap, brands have poured significant investments into a sprawling middle layer of stand alone customer data platforms (CDPs) and reverse ETL tools.

This is an expensive “middleware tax” that exposes brands to three critical infrastructure failures:

• Context decay: When data architecture relies on middleware, customer data lags significantly. In high-velocity industries like eCommerce, media, or travel, a half-hour delay means you aren’t messaging a live customer—you are messaging a ghost of their past behavior. By the time an email or a mobile push notification drops, the moment has passed.
• The AI-data divide: Many companies host their proprietary and open source foundation models on cloud data lakehouses like Databricks, but those models end up trapped as “shelfware.” Marketers often don’t have an easy way to integrate and leverage those models within their agentic workflows.
• The financial attribution black box: Because customer engagement platforms and their data warehouses are often disconnected, calculating the exact financial impact of a campaign can be a nightmare. Without a direct link between a downstream engagement click and an upstream transaction in the warehouse, leaders see a black box instead of a clear picture of campaign ROI.

To fix this experience, we have to fix the architecture.

Today, Braze is thrilled to announce a broader bi-directional integration with Databricks to eliminate the middleware tax and deliver a composable solution for customer engagement. This partnership leverages Databricks CustomerLake for unified audience management, Databricks Model Serving for intelligent agentic workflows, and Delta Sharing for real-time data synchronization. By pairing Databricks as your source of enterprise data with the power of the Braze platform, this integration directly bridges the gap between data and engagement.

"CustomerLake represents a real shift in how brands can act on their data, and Braze is proud to be at the center of it. By embedding directly into Databricks, we're giving customers the ability to move from insight to engagement without the friction, latency, or cost of middleware. The brands that win today are the ones that can meet their customers in the moment. That's exactly what this makes possible."

Ed McDonnell

Chief Revenue Officer, Braze, Braze

Here’s how our joint solution helps our shared customers.

Lakehouse-native, zero-copy control

We believe data teams should never have to compromise governance, introduce vendor risk, or replicate massive datasets into third-party silos just to give marketing teams the agility they need. By integrating the engagement power of Braze with your enterprise customer intelligence in Databricks, you can significantly reduce the middleware tax altogether. This lakehouse-native architecture helps sensitive customer profiles remain securely under your control, while your Marketing team gains real-time access to the production-grade data required to fuel real-time engagement without the complexity of custom ETL pipelines.

Braze customers today already have the ability to sync relevant user data from Databricks using Braze Cloud Data Ingestion (CDI). With the launch of Databricks CustomerLake, an Agentic CDP embedded natively within Databricks, customers can now push their governed cohort audiences into Braze for segmentation to help marketers deliver the next generation of personalized customer experience.

"The biggest bottleneck in enterprise marketing is getting the right data. The new Braze Data Platform enhancements, including the bidirectional Databricks integration, fundamentally change that by solving the data handoff problem at the platform level. Marketers can now act on signals the same day they exist, not the same quarter. Stitch has been doing this work with enterprise brands across retail, streaming, and healthcare, and the limiting factor has consistently been the data handoff problem. That problem is now solved at the platform level. We’re proud to be the only services partner named in both the Braze and Databricks CustomerLake launches because we’ve seen firsthand what this unlocks for our clients."

Michael Burton

CEO of Stitch, Stitch

These capabilities eliminate the middleware tax entirely by removing the need for standalone CDPs or reverse ETL tools. It significantly lowers your total cost of ownership (TCO) and minimizes engineering debt, freeing data teams from the exhausting cycle of writing, monitoring, and repairing brittle data pipeline scripts.

Governed models for agentic workflows

We recently launched the BrazeAI Agent ConsoleTM to give marketers the ability to create, configure, and deploy agents flexibly within their customer journeys. These agents can generate content, make intelligent decisions, and enrich data so marketers can deliver more personalized customer experiences. These agents use large language models (LLMs) to interpret inputs, generate responses, and perform reasoning. As part of this capability, we give customers the option to use a Braze-provided LLM or to connect their own foundation model providers such as OpenAI, Anthropic, or Google Gemini.

To give customers even more flexibility with Agent Console, we’ve now made it possible for them to leverage Databricks Model Serving, a library of proprietary and open source models hosted by Databricks, so they can automate advanced conversational workflows and drive high-quality, hyper-personalized, 1-to-1 customer lifecycles at scale. This also enables them to improve governance and the cost control benefits of centralizing LLM usage and cost management directly within Databricks.

Bi-directional feedback loop

Customer context changes by the sub-second, and your data architecture needs to reflect that reality. True real-time engagement requires a continuous ecosystem where live marketing interactions immediately refresh the central lakehouse, ensuring that marketers don’t message the ghost of past behavior and data teams have complete visibility into business performance.

This feedback loop is now fueled by Databricks Delta Sharing, the most widely adopted open protocol for secure data sharing. Delta Sharing securely and continuously streams granular, real-time engagement data, including message sent, opened, clicked, and bounced, directly back into your Databricks environment.

Marketers can execute high-velocity, time-sensitive campaigns like cart abandonment triggers, price drops, or real-time travel alerts before the window of opportunity slams shut. Concurrently, data teams can join live campaign logs with actual transactional tables, giving CFOs lakehouse-validated financial attribution to help prove the ROI of their campaigns.

Final thoughts

Modern consumers do not experience your brand in batch syncs or scheduled data uploads, they experience it in the moment. When your infrastructure relies on clunky middleware and slow data pipelines, your customers are the ones who feel the friction through outdated messages and missed opportunities.

By bridging the gap between the enterprise scale of Databricks and the real-time orchestration and engagement of the Braze platform, you change the way your brand interacts with the world. The result? Less noise, deeper personalization, and a clear, lakehouse-validated view of your marketing ROI.

Take control of your data and your customer experience. Get in touch with our team today to learn how to unlock the true value of your data science and customer engagement investments.