CustomerLake in One Sentence: Marketing Moves Into the Lakehouse
Databricks CustomerLake is an agentic CDP platform that runs customer data management, identity resolution, AI models, and always-on personalization directly on a customer data lakehouse instead of a separate marketing system, so decisions and activation happen where the data already lives. Databricks has moved beyond data infrastructure and into marketing software by launching CustomerLake, an “agentic” customer data platform designed to manage customer data and run AI-driven marketing campaigns inside the same environment. The product was announced at the company’s recent Data + AI Summit and is currently in Private Preview, with early testing underway at brands like HP, Circle K, AB InBev, and Getnet by Santander. This is not another UI on top of a warehouse; it is a direct attack on the idea that marketing needs a standalone CDP at all.
From Siloed CDPs to a Customer Data Lakehouse
CustomerLake unifies customer data, AI models, agents, identity resolution, audience building, and activation inside Databricks, collapsing what used to be separate CDP and orchestration layers into the lakehouse. The core idea is blunt: reduce the distance between where customer data lives and where marketing actions get executed. That makes CustomerLake less a classic CDP and more a customer data lakehouse with execution built in. Circle K’s early use case—building audiences and activating them via Adobe while avoiding moving its full data lake into another environment—shows how this design cuts out duplication, reconciliation, and slow batch exports. In other words, CustomerLake turns “data gravity” from a constraint into the design principle. If your system of record and your system of decisioning are the same thing, traditional CDP middleware looks like unnecessary routing.
Agentic CDP: Always-On Personalization Without Campaign Drudgery
The real break with past CDPs is the agentic approach. CustomerLake uses agents to analyze customer behavior and decide the offer, message, channel, and timing for each individual in real time, turning discrete campaigns into continuous loops. Databricks describes these “infinity campaigns” as loops that constantly analyze and act, pushing marketers toward always-on personalization where the bottlenecks are data access, model execution, and controls, not manual campaign setup. One analysis calls this a ground-up build of AI native marketing technology that offers agents for both data handling and customer engagement, explicitly advocating a shift from traditional campaigns to always-on, continuous engagement. That matters: instead of orchestrating segments and journeys step-by-step, marketers move to supervising a system that drafts briefs, builds audiences, resolves identities, and deploys campaigns as an ongoing process.
Why Infrastructure-Native Marketing Is the Direction of Travel
CustomerLake builds directly on existing Databricks data infrastructure, promising better data economics, more consistent data use, and closer alignment with enterprise IT strategy. Circle K’s experience highlights why this timing matters: moving large, governed datasets into separate martech environments has become a serious friction point; executing where the data already lives cuts time-to-activation and the pain of data movement. One commentary argues that CustomerLake fast-tracks trendlines toward embedded capabilities and functional expansion, and shows the power of AI-enabled software engineering to accelerate martech innovation. Databricks also makes a pointed category argument: if data, models, identity, and activation sit near the warehouse, some CDP “middleware” functions may collapse into the core data platform. In plain terms, this is AI-native marketing infrastructure threatening the long-term relevance of standalone CDP vendors and even some marketing clouds.
What Marketers Should Do Before CustomerLake Hits General Release
CustomerLake is still in private preview and is expected to reach general availability later in 2026, making current deployments a live test of enterprise appetite for agentic AI in marketing. It already includes AI-driven identity resolution, access to third-party identity graphs, and an open activation ecosystem spanning major ad and engagement platforms. But Databricks ties this power to Unity Catalog governance and a “humans in the loop” path—acknowledging that agentic execution demands clear policies on what agents can do automatically, what needs approval, and what audit trails must exist. The conclusion is uncomfortable for legacy CDP vendors: as AI-native, infrastructure-first tools like CustomerLake enter the marketing automation space, classic batch-era CDPs risk becoming sidecars. If your customer strategy still depends on nightly exports and calendar-driven campaigns, the change is not optional; it is overdue.
