metrics-layer

Source-available semantic layer and sidecar for agentic AI and governed analytics. Compiles declarative YAML models into optimized SQL, KPIs, and semantic context across 8 SQL dialects.

A DuckDB extension implementing Measures in SQL

Archived. Earlier Bonnard product. Current: bonnard.dev / @bonnard/mcp-charts

The open source metrics layer

Archived. Earlier Bonnard product. Current: bonnard.dev / @bonnard/mcp-charts

Agentic BI: Semantic and Ontology-driven Data Agent for natural language data Q&A，analysis，insights，and action recommendations.

Semantic layer / dbt package for quantitative Product-Market Fit (PMF) analytics — Growth Accounting, Cohort Analysis, and Distribution (Tribe Capital framework).

Production-style semantic metrics governance platform with metric contracts, lineage, trust scoring, conflict detection, and AI-agent readiness validation.

Deterministic cohort compiler in Rust. Refuses to answer ambiguous questions: when a business noun maps to several defensible predicates, it makes you name which one.

AI-powered semantic layer for BI: auto-generates metrics definitions, dimension hierarchies, certified datasets and natural language descriptions from raw BigQuery schemas using LLMs

A tested metric layer over 1M real retail transactions: every KPI defined once in YAML with its judgement calls attached, and a generated dashboard

Kimball star schema in DuckDB where every executive KPI is defined once in a YAML metric contract, compiled to DuckDB and Snowflake SQL with sqlglot. A CI gate runs each contract against the dashboard measure that claims to implement it and fails the build on drift, naming the root cause: missing filter, grain fan-out or wrong measure basis.

Governed NLQ over a locked metric catalogue: the model picks a metric, it never writes the SQL or computes the number. Every answer is cited; ambiguous questions get a question back.