According to an announcement by Salesforce, its collaboration with Snowflake under the Zero Copy model is being expanded with the goal of turning distributed enterprise data into trusted context for humans and AI agents, without requiring customers to move their data from the platforms where it is stored. According to the source, the AI model itself is no longer the differentiator it once was, as enterprises increasingly have access to the same powerful foundation models. What will separate leading companies is their unique business context—including customer relationships, transactions, product usage, conversations, documents, business definitions, and institutional knowledge. This context is not centralized in a single location, but spans across Salesforce, data platforms such as Snowflake, applications, documents, and other systems, which is why the Data 360 platform was developed.
From Enterprise Data to Business Context and the Agent Context Engine
At the Dreamforce conference, Salesforce showcased how Data 360 works in collaboration with Snowflake as a Zero Copy partner to advance Data 360 as a context engine for the agentic enterprise. According to the publication, raw data alone is not sufficient for AI operations; an agent needs to understand what the data means, how different pieces of information relate to each other, what it must remember, what data it is permitted to access, and what actions it can safely take. The source describes this evolution as a gradual progression: from data to semantics, from there to context, and finally to action (data → semantics → context → action).
At the core of the Data 360 platform is the Agent Context Engine (ACE), which brings together structured data, unstructured content, knowledge graphs, and agent memory into a unified context layer. The ACE layer provides agents with real-time context in the structure they require to reason more effectively across different channels and large language model (LLM) providers, while helping improve grounding, governance, latency, and efficiency. Through a bidirectional Zero Copy approach to Snowflake, this context layer can incorporate governed enterprise data directly from where it is stored, without creating an additional copy of the information.
Extending Data 360 to an Agentic Customer Data Platform and External Systems
According to the company's announcement, Data 360 extends trusted context to additional use cases, applications, and agents. Within this framework, the Customer Data Platform (CDP) is defined as an Agentic Customer Data Platform, integrating agents across Data 360, including goal-based segmentation. This integration evolves the CDP from a system that organizes customer data into one that helps businesses put data to work, enabling teams to move from defining business goals to creating precise audiences and activating them across various channels.
Concurrently, the Data 360 Headless capability extends trusted context beyond Salesforce applications. This capability allows developers to bring customer data and shared semantic definitions into custom applications, external tools, and third-party agents via open standards, including the Model Context Protocol (MCP), APIs, and skills.
Data Sharing Without Duplication via the v2 Connector and Iceberg Tables
For customers holding data in Snowflake, the Zero Copy v2 connector removes the last major friction in how joint customers share data. Teams building AI solutions on Snowflake using Cortex AI gain access to fresh Salesforce data inside Snowflake without the need to build traditional ETL data pipelines.
Under this architecture, objects from Data 360 appear as native Apache Iceberg tables inside the customer's Snowflake account. These tables can be queried using standard SQL, are compatible with Streams and Dynamic Tables for building near real-time change data capture (CDC) pipelines, and are configured as AI-ready out of the box. The v2 connector allows teams to use Snowflake scheduled tasks to capture changes without maintaining an additional pipeline, while authentication via OpenID Connect (OIDC) helps secure access to this data.
Data Governance, Multicloud Connectivity, and Open Protocols
In addition to the data connector, the expanded collaboration includes integration with Horizon Catalog, enabling direct file federation with Snowflake-managed Apache Iceberg tables. This configuration allows Data 360 to query governed data and use it to ground AI experiences without moving the underlying information. The integration brings the governance and metadata mechanisms of both platforms closer together, aiming to maintain consistent policies in data usage across analytics, applications, and AI.
At the infrastructure level, the Data 360 Private Connect Multicloud solution for Snowflake on Azure allows organizations to privately connect Data 360 running on AWS with Snowflake running on Azure, without network traffic traversing the public internet—a capability of importance for customers in highly regulated environments. Alongside this, open communication protocols including MCP-based data management servers and cross-platform integrations are designed to enable Snowflake agents and Salesforce agents to securely access shared enterprise context and execute governed workflows with full auditability and control. Furthermore, Tableau will natively consume semantic views authored in Snowflake, enabling organizations to define business metrics once and use them consistently across dashboards, Cortex queries, and Agentforce workflows.
Enterprise Deployment Metrics, Customer Testimonials, and Availability Timelines
According to metrics presented at Dreamforce, where Snowflake was recognized as the 2026 System of Context Partner, more than 1,000 joint customers are currently running workloads across both platforms using native Zero Copy, with 71 trillion connected rows—a figure representing more than 4× year-over-year growth.
Saptarshi Mukherjee, Senior Director of Product Management, noted that without governed context and business semantics, agents risk producing high-confidence answers based on incorrect assumptions and inconsistent definitions, and in collaboration with Salesforce they are solving this critical bottleneck by tightly grounding agents in trusted data where it already lives. Mike Doll, VP of Data at Guitar Center, stated that the bidirectional Zero Copy integration felt like a 1,000× step forward versus their old setup, transforming how they serve musicians across Guitar Center, and that building on this foundation, these new integrations should help bring trusted enterprise context directly to AI agents and customer workflows. Ranjith Ramesh, Senior Engineering Manager for GTM Intelligence at Indeed, added that through this integration, the company built a real-time connected data layer powering demand routing, audience activation, and their internal AI platform.
Regarding feature availability, the Zero Copy Data Sharing v2 connector and Data 360 Private Connect Multicloud for Snowflake on Azure are Generally Available. The Zero Copy extension with Horizon Catalog and Snowflake Cortex Agents is available in Public Preview. Salesforce and Informatica MCP servers in Snowflake Cortex are in Private Preview on the roadmap, and Tableau semantic views for Snowflake are planned for General Availability in early 2027.