Rill と ClickHouse: メータリング時代のリアルタイム運用 BI

Rill is an operational BI-as-code tool designed to unify and analyze massive real-time event data generated by AI systems and cloud costs, processing over 100 billion events daily using ClickHouse as its core architecture. By defining data ingestion and metrics with SQL and YAML, Rill allows for version control through Git and facilitates fast development with AI tools like Cursor, enabling precise conversational analysis. The tool addresses the challenge of understanding interconnected data by integrating operational and financial data to provide comprehensive business insights. Founded by Mike Driscoll and Nishant Bangarwa after their experience with Metamarkets, Rill's architecture keeps analytics close to the database, using dlt for declarative data loading and ClickHouse for high-performance aggregation. The system's architecture ensures that business logic and metrics are consistently applied across dashboards, APIs, and programs, supporting real-time interaction with live data. This approach shifts analytics from retrospective reporting to operational decision-making, treating analytics as an integral part of the business infrastructure.