Snowflake develops semantic layer to power enterprise AI coworkers

Snowflake is developing an internal semantic layer to address data inconsistency within large enterprises. This layer acts as an intermediary that translates governed business language into physical database schemas, creating a “golden layer” of data meaning. By providing standardized semantic views, the company aims to ensure that both humans and AI agents receive consistent answers to business questions, such as defining an “active customer.”

This technical shift aligns with a broader industry trend where providers are rebranding their AI offerings from “intelligence” tools to “enterprise AI coworkers.” Snowflake has rebranded its Intelligence in CoWork, while Databricks has introduced Genie One as an “agentic coworker.”

Unlike previous analytical tools that required users to visit a specific interface to ask questions, these new “coworker” models focus on integration. By utilizing connectors for platforms like Microsoft Teams, Excel, and Google Sheets, these AI agents can perform tasks—such as drafting documents or creating reports—directly within the user's existing workflow, rather than just providing answers in a standalone application.