Skip to main content

Core concepts

The semantic layer​

Celiq sits between your database and your analysts. Instead of writing SQL every time, you define your data model once in Forge using CML (Celiq Modelling Language) — then anyone can query it without writing SQL.

Database tables
     ↓
  Nodes  ← define fields, metrics, joins
     ↓
 Domains ← group nodes into a business context
     ↓
 Reveals ← curated views with pre-selected fields
     ↓
Discover ← drag-and-drop query builder + 27 chart types
     ↓
 Mosaics ← dashboards of pinned tiles

Nodes​

A node maps to one database table. It defines:

  • Attributes — dimensions (strings, dates, IDs)
  • Metrics — measures (SUM, COUNT, AVG, etc.)
  • Joins — relationships to other nodes

Example: payment.node.celiq

node: payment
table: public.payment
attributes:
  - name: payment_id
    type: integer
    role: primary_key
  - name: payment_date
    type: date
    is_date: true
metrics:
  - name: total_revenue
    label: Total payment revenue
    sql: SUM(amount)
    type: currency

Domains​

A domain groups related nodes into a business context. It defines joins between nodes and contains one or more reveals.

Example: pagila.domain.celiq

Reveals​

A reveal is a curated view inside a domain. It pre-selects the most useful fields, suggests metrics, and optionally sets default filters. This is what analysts pick in Discover.

Mosaics​

A Mosaic is a dashboard made of tiles. Each tile is pinned from a Discover query and runs live against your warehouse.

Orion​

Orion is Celiq's built-in AI assistant. It can:

  • Generate node and domain CML from your database schema
  • Answer questions in natural language ("What was total revenue last month?")
  • Create Mosaics and WBR reports
  • Explain metric definitions and SQL

CML​

CML (Celiq Modelling Language) is YAML-based. Node files end in .node.celiq, domain files in .domain.celiq. Files are stored in your project's Git repository for version control.