How CustomerLake's Agentic Architecture Works

CustomerLake replaces traditional, manual campaign workflows with two autonomous agent types that work continuously :

Profile Agents automatically assemble fragmented raw customer data from sources like CRM, transactions, and web behavior into unified, business-ready Customer 360 profiles. They handle identity resolution and data unification without manual intervention, giving marketers a single, governed view of each customer .

Campaign Agents run what Databricks calls "infinity campaigns" — always-on, autonomous marketing programs rather than static, time-boxed campaigns. They build audiences, recommend next-best actions, activate across channels, and continuously optimize based on real-time performance . The agents use lakehouse data and AI models to decide what content to send, when to send it, and to which audience, closing the loop between customer data and execution .

This agentic workforce represents a fundamental shift from platforms that suggest actions to ones that execute and optimize autonomously within bounded rules .

Launch Partners and Early Adopters

Databricks introduced CustomerLake with a broad ecosystem of launch partners that provide critical capabilities on top of the platform:

• Adstra provides its Conexa identity resolution platform as an identity layer inside CustomerLake, allowing brands to build trusted customer profiles without moving data out of the governed environment .
• Bloomreach extends its Loomi agentic personalization platform across email, web, and more through native integration, creating a direct path from unified data to real-time personalized execution .
• Iterable achieved Validated Partner status and joined as a launch partner, providing the orchestration layer that animates CustomerLake's agentic intelligence through personalized messaging .
• Acxiom powers identity resolution through its Real ID service, enabling AI-ready marketing on the platform .
• Slalom serves as a services launch partner, helping clients bring together customer data, AI, and marketing activation on a trusted foundation .

Early adopter customers presenting at the summit included HP, which is building an AI-ready B2B go-to-market data infrastructure, and Circle K, which is powering personalized loyalty programs with customer data at scale . Other enterprise customers mentioned include AB InBev and Getnet by Santander .

The wider partner ecosystem for data ingestion and activation spans Adobe, Meta (Audience and Conversions API), Epsilon, LiveRamp, The Trade Desk, and others .

Why CustomerLake Reflects the Agentic Enterprise Shift

CustomerLake's architecture and launch timing align with three converging market trends that are reshaping how enterprises buy and use software:

1. From AI Assistants to Autonomous Execution

The marketing technology stack is pivoting rapidly from tools that merely suggest or assist to autonomous systems that plan, act, and optimize across channels with minimal human intervention . Gartner projects that by the end of 2026, 40% of enterprise applications will include embedded agentic AI, up from under 5% in 2025 . A separate prediction suggests that 25% of enterprise software purchases will include an embedded agent component by Q4 2026, meaning buyers will expect agentic capability inside existing platforms like CRM, ERP, and marketing suites rather than purchasing standalone agent tools .

2. The CDP Category Is Being Re-Architected Around Agents

Traditional CDPs export static audience lists to separate execution tools, creating latency and data duplication. CustomerLake embeds agentic capabilities directly into the data lakehouse, meaning customer data never has to move or be copied to be activated . This approach reflects a broader industry view that the era of static audience segments is giving way to platforms that govern, feed, and orchestrate AI agents in real time . Industry observers note that the question enterprises should now ask when evaluating a CDP has changed: it is no longer enough to ask whether it unifies data — the key question is whether it is ready to feed autonomous agents .

3. Governance as the New Competitive Moat

By keeping marketing agents inside the governed lakehouse, Databricks addresses a persistent pain point that has limited enterprise adoption of AI in marketing: data duplication, security fragmentation, and lag between analysis and activation. The bet is that enterprises will prefer to activate agents where their sensitive customer data already lives, under existing governance controls, rather than moving it to a separate marketing cloud . This makes the platform's governance layer — rather than any single AI model — the true competitive moat.

CustomerLake signals that Databricks views agentic marketing not as a feature but as a platform inflection point. The broader contest in enterprise software is increasingly about which platform becomes the default runtime for autonomous business agents, and CustomerLake is Databricks' opening move in that race .

How CustomerLake's Agentic Architecture Works

CustomerLake replaces traditional, manual campaign workflows with two autonomous agent types that work continuously :

Profile Agents automatically assemble fragmented raw customer data from sources like CRM, transactions, and web behavior into unified, business-ready Customer 360 profiles. They handle identity resolution and data unification without manual intervention, giving marketers a single, governed view of each customer .

Campaign Agents run what Databricks calls "infinity campaigns" — always-on, autonomous marketing programs rather than static, time-boxed campaigns. They build audiences, recommend next-best actions, activate across channels, and continuously optimize based on real-time performance . The agents use lakehouse data and AI models to decide what content to send, when to send it, and to which audience, closing the loop between customer data and execution .

This agentic workforce represents a fundamental shift from platforms that suggest actions to ones that execute and optimize autonomously within bounded rules .

Launch Partners and Early Adopters

Databricks introduced CustomerLake with a broad ecosystem of launch partners that provide critical capabilities on top of the platform:

• Adstra provides its Conexa identity resolution platform as an identity layer inside CustomerLake, allowing brands to build trusted customer profiles without moving data out of the governed environment .
• Bloomreach extends its Loomi agentic personalization platform across email, web, and more through native integration, creating a direct path from unified data to real-time personalized execution .
• Iterable achieved Validated Partner status and joined as a launch partner, providing the orchestration layer that animates CustomerLake's agentic intelligence through personalized messaging .
• Acxiom powers identity resolution through its Real ID service, enabling AI-ready marketing on the platform .
• Slalom serves as a services launch partner, helping clients bring together customer data, AI, and marketing activation on a trusted foundation .

Early adopter customers presenting at the summit included HP, which is building an AI-ready B2B go-to-market data infrastructure, and Circle K, which is powering personalized loyalty programs with customer data at scale . Other enterprise customers mentioned include AB InBev and Getnet by Santander .

The wider partner ecosystem for data ingestion and activation spans Adobe, Meta (Audience and Conversions API), Epsilon, LiveRamp, The Trade Desk, and others .

Why CustomerLake Reflects the Agentic Enterprise Shift

CustomerLake's architecture and launch timing align with three converging market trends that are reshaping how enterprises buy and use software:

1. From AI Assistants to Autonomous Execution

The marketing technology stack is pivoting rapidly from tools that merely suggest or assist to autonomous systems that plan, act, and optimize across channels with minimal human intervention . Gartner projects that by the end of 2026, 40% of enterprise applications will include embedded agentic AI, up from under 5% in 2025 . A separate prediction suggests that 25% of enterprise software purchases will include an embedded agent component by Q4 2026, meaning buyers will expect agentic capability inside existing platforms like CRM, ERP, and marketing suites rather than purchasing standalone agent tools .

2. The CDP Category Is Being Re-Architected Around Agents

Traditional CDPs export static audience lists to separate execution tools, creating latency and data duplication. CustomerLake embeds agentic capabilities directly into the data lakehouse, meaning customer data never has to move or be copied to be activated . This approach reflects a broader industry view that the era of static audience segments is giving way to platforms that govern, feed, and orchestrate AI agents in real time . Industry observers note that the question enterprises should now ask when evaluating a CDP has changed: it is no longer enough to ask whether it unifies data — the key question is whether it is ready to feed autonomous agents .

3. Governance as the New Competitive Moat

By keeping marketing agents inside the governed lakehouse, Databricks addresses a persistent pain point that has limited enterprise adoption of AI in marketing: data duplication, security fragmentation, and lag between analysis and activation. The bet is that enterprises will prefer to activate agents where their sensitive customer data already lives, under existing governance controls, rather than moving it to a separate marketing cloud . This makes the platform's governance layer — rather than any single AI model — the true competitive moat.

CustomerLake signals that Databricks views agentic marketing not as a feature but as a platform inflection point. The broader contest in enterprise software is increasingly about which platform becomes the default runtime for autonomous business agents, and CustomerLake is Databricks' opening move in that race .