rilldata/rill

What it solves

Rill provides a high-performance business intelligence (BI) tool designed for both humans and AI agents. It solves the problem of slow, manual BI workflows by treating BI-as-code, allowing developers to define models, metrics, and dashboards in YAML and SQL, which can then be authored and managed by coding agents.

How it works

Users connect their data via connectors (S3, GCS, databases) and use OLAP engines like ClickHouse or DuckDB for sub-second query performance. The system uses a semantic layer defined in YAML to create a single source of truth for dimensions and measures, which generates SQL at query time. This architecture allows for conversational BI where users can query metrics in natural language and AI agents can connect directly to the semantic layer via an MCP server.

Who it’s for

It is designed for developers and business users who need real-time, embeddable analytics dashboards and the ability to integrate AI agents into their data exploration and reporting workflows.

Highlights

• BI-as-code: Projects are defined using YAML and SQL, making them compatible with AI coding agents like Claude Code and Cursor.
• Semantic Layer: Centralized definition of metrics and dimensions to ensure consistency across queries.
• Agent-First: Includes an MCP server to connect AI agents (Claude, ChatGPT) directly to business metrics.
• High Performance: Powered by OLAP engines like ClickHouse and DuckDB for sub-second queries at scale.
• Embeddable: Dashboards and APIs can be integrated directly into other products.