With the rapid growth of cloud data warehouses, a new ecosystem of tools has emerged, including transformation, testing, quality, observability, and vertical-specific reporting tools. This boom necessitates a new generation of business intelligence: headless BI, which decouples data metrics from presentation layers to push metrics definition up the data stack. A standardized definition of headless BI includes four essential components: data modeling, access control, caching, and APIs. Data modeling involves defining metrics in code, ensuring deduplication, and consistency across applications. Access control manages and audits access to data, while caching ensures consistent data freshness and reduces redundant requests. APIs make data accessible to every application, enabling new automations and use cases. A headless BI system is valuable when integrated into any data stack and should be made available to everyone, with community input and values guiding its stewardship.