← All patterns Martech Patterns · September 22, 2026

The Data Layer Race · Whoever Holds the Data Controls the Agents

Analysis by the CDP Institute

AI agents are starting to work directly off a company's live data instead of a copy, and every layer of the data stack — the platforms that move and clean it, the customer-profile layer over the warehouse, the tools that fetch it — is angling to be the one they sit on. Whoever becomes that gatekeeper decides what an agent can see and how far it can go, so control is shifting to the data layer, not the apps built on top.

  1. Dbt Labs–Fivetran Merger Unifies Data Integration and Transformation for AI-Focused Partners

    What happened: Dbt Labs and Fivetran have merged into a single, yet-unnamed company and are rolling out joint channel plans focused on AI data infrastructure that combines Fivetran’s data integration with dbt’s transformation capabilities.

    Why it's important: By unifying extraction and transformation under one commercial and technical umbrella, the merged company can standardize how semantic models, lineage, and access policies flow into AI-facing data products. This shifts key governance and observability decisions upstream into the ELT layer, where partners and enterprises will now decide how safely to expose unified profiles to agents across clouds and tools.

    Source: CRN

  2. AI Customer Insights Agent from Syntasa Debuts on Google Cloud Marketplace

    What happened: Syntasa launched its Customer Insights Agent on Google Cloud Marketplace, offering an AI assistant that queries and activates customer data on Google Cloud infrastructure.

    Why it's important: Embedding an AI insights agent inside the cloud data platform shifts customer analytics from exported datasets to live warehouse-native profiles, forcing teams to decide how to enforce governance, observability, and guardrails at the agent–data boundary instead of in downstream tools.

    Source: syntasa.com

  3. Antfly Raises Funding To Ease AI Data Retrieval

    What happened: Antfly raised $2 million to develop infrastructure that reduces the operational burden on developers building custom data retrieval for AI agents.

    Why it's important: By offloading bespoke retrieval plumbing, organizations can spend more time shaping what data agents should access and less on how to connect it.

    Source: geekwire.com

From the CDP Institute: What is a CDP? · Vendor Directory · News archive