Basedash Semantic Layer

Product Introduction

1. Definition: The Basedash Semantic Layer is an enterprise metadata management and metrics layer within the Basedash AI-native business intelligence platform. It is a centralized system for defining, storing, and managing reusable SQL metrics and data models that serve as a governed "source of truth" for analytics.
2. Core Value Proposition: It solves the critical problem of metric inconsistency and redundant code by allowing data teams to define metrics once and have them referenced deterministically by all AI-powered components—chat, charts, dashboards, insights, and automations—ensuring consistent calculations across the entire data stack.

Main Features

1. Deterministic SQL Definitions: This is the core technical foundation. Users create saved SQL queries scoped to specific data sources. Each definition includes a human-readable name, a machine reference name, a descriptive explanation, and the exact SQL query. This transforms ad-hoc SQL into version-controlled, AI-readable analytical building blocks. For example, the activation_rate metric is defined with precise SQL logic filtering users who onboard within seven days, eliminating ambiguity.
2. AI-Aware Metric Catalog: The semantic layer automatically provides Basedash AI agents with a catalog of available definitions. When an AI chat query is made or a chart is generated, the AI can inspect these definitions, understand their logic, and reference them in generated SQL using Liquid syntax (e.g., {{ definition("mrr") }}). This enables AI to perform complex, metric-aware analyses while adhering to pre-approved business logic.
3. Universal Reuse Across BI Workflows: Definitions are integrated into every AI-driven workflow. They can be referenced inside Common Table Expressions (CTEs) within the SQL editor for complex queries, ensuring dashboards display the same calculation as a chat answer or an automated insight report. This enforces consistency and prevents "metric drift" where different reports show conflicting numbers for the same KPI.
4. Composable SQL Modeling with Liquid Syntax: The system uses Liquid templating syntax to enable modular SQL composition. Teams can build larger analytical queries by referencing reusable definitions. This promotes clean, maintainable code by avoiding the duplication of complex metric logic across multiple dashboards and reports, significantly reducing technical debt in the analytics codebase.

Problems Solved

1. Pain Point: Inconsistent Metrics and Divergent Dashboards. Organizations struggle with "metric anarchy," where different teams, dashboards, or reports calculate the same KPI (e.g., Monthly Recurring Revenue, Churn Rate) using slightly different SQL logic, leading to conflicting numbers and eroded trust in data. The semantic layer provides a single source of truth to eliminate this.
2. Target Audience: This product is essential for Data Analysts, Analytics Engineers, and BI Developers who build and maintain dashboards and reports. It is also critical for Data Scientists and AI Engineers who need to leverage trusted business metrics in automated models and insights. Product Managers and Marketing Managers who consume data benefit from the consistency it guarantees.
3. Use Cases:
   • Standardizing KPI Reporting: A finance team defines mrr (Monthly Recurring Revenue) once, excluding trials and one-time credits. Every revenue dashboard, board deck chart, and AI-generated revenue insight references this single, approved definition.
   • Scaling AI-Powered Analytics: A user asks the AI chat, "What was our activation rate last week?" The AI uses the semantic layer to fetch the activation_rate definition, executes the precise SQL, and returns a trustworthy answer consistent with all other reporting.
   • Building Consistent Automated Reports: An automated daily briefing that uses the cohort_retention definition to analyze user engagement will always use the same cohort logic as the weekly retention dashboard, ensuring trend comparisons are valid.
   • Onboarding and Governance: New analysts can quickly understand core business metrics by reading the centralized definitions in the catalog, accelerating their time-to-insight and ensuring they use approved calculations from the start.

Unique Advantages

1. Differentiation vs. Traditional BI and Standalone Metrics Layers: Unlike traditional BI tools where metric logic is often locked within individual dashboard components, the Basedash Semantic Layer is intrinsically linked to a full AI-native suite. It differentiates itself from generic metrics stores (like a simple dbt model catalog) by being natively consumed by the platform's AI agents for chat, chart generation, and insights. The semantic layer is not just a repository; it's an active, utilized component of the analytics workflow.
2. Key Innovation: AI-Native Metric Utilization. The pivotal innovation is the tight integration between the defined metrics and the AI capabilities. The AI is explicitly given a catalog of definitions and instructed to use them. This creates a feedback loop where deterministic, human-defined metrics steer probabilistic AI outputs, combining the reliability of governed SQL with the speed and flexibility of AI interaction. This ensures that the "chat-to-chart-to-insight" pipeline is built on a foundation of trusted, consistent business logic.

Frequently Asked Questions (FAQ)

1. How is the Basedash Semantic Layer different from a metrics layer like MetricFlow or dbt metrics? The key difference is the native AI integration. While solutions like dbt metrics define semantic models, the Basedash Semantic Layer is purpose-built to be read and referenced by AI agents within the same platform. It turns metrics into not just SQL definitions, but into context for AI to understand and use when answering questions, building charts, and generating automated insights, providing a seamless end-to-end analytics experience.

2. Can I reference a Basedash definition within a custom SQL query or CTE? Yes. The system uses Liquid syntax for composable SQL. Within the Basedash SQL editor or a dashboard query, you can reference any saved definition using {{ definition("definition_name") }}. Best practice is to place these references inside a CTE (Common Table Expression) for clarity. For example: WITH mrr AS ({{ definition("mrr") }}) SELECT * FROM mrr WHERE....

3. What specific technologies does the Basedash Semantic Layer use? It is built on a combination of a metadata repository for storing definitions, a SQL execution engine to run queries against connected data warehouses (supporting 750+ data sources), and a Liquid template engine for dynamic SQL composition. The AI layer utilizes the stored definition metadata (name, description, SQL) to make decisions on metric usage.

4. How do I migrate our existing SQL metrics into the Basedash Semantic Layer? Basedash offers support for migration from other BI tools as part of their onboarding. The process involves identifying core KPIs (like activation rate, retention, revenue metrics), translating their logic into the Basedash definition format (name, reference name, description, SQL), and saving them to the appropriate data source catalog. This centralizes your metrics and makes them immediately available for all AI workflows.

5. What is the difference between a "definition" in the semantic layer and a "skill" in Basedash? They serve complementary purposes. Definitions (semantic layer) are for deterministic SQL calculations—the exact numerical computation of a metric like MRR. Skills are prose instructions for the AI about broader business context, goals, or analytical approaches (e.g., "Always compare activation rates on a weekly basis and note major marketing campaigns"). Use definitions for the "what" (the calculation) and skills for the "how" (the analytical guidance).