Product Types Overview
Qarion supports a rich, extensible system of data product types designed to catalog the full spectrum of AI, data, and content assets. Every asset in the catalog is a Data Product, and its type determines which specialized metadata, tabs, and workflows are available.
How Product Types Work
Product types operate on three layers:
| Layer | What it does | Who manages it |
|---|---|---|
| System Types | Built-in types used by sync connectors (dbt, scrapers, BI tools) | Platform (automatic) |
| Default Types | Pre-seeded types available in every new space | Platform (automatic) |
| Custom Types | User-defined types with custom icons, colors, and workflows | Space administrators |
When you create or import a data product, its type is either set automatically by a connector (e.g., dbt sync creates "Table" or "View" products) or chosen manually from the space's type menu.
Type Categories
Qarion organizes its product types into five categories, each designed for different kinds of data assets:
🗄️ Data Infrastructure
Traditional data warehouse, BI, and pipeline assets.
| Type | Description | Typical Source |
|---|---|---|
| Table | Database table or view | dbt sync, Scraper |
| Dashboard | Analytics dashboard or report | BI connector, Manual |
| API | REST or GraphQL API endpoint | Manual |
| Data Stream | Real-time data stream or Kafka topic | Manual |
| File/Dataset | File or dataset (CSV, Parquet, etc.) | Manual |
| Notebook | Jupyter or analysis notebook | Manual |
These types get the full universal feature set: schema, profiling, quality checks, lineage, governance, contracts, and more. Learn more →
🤖 AI & Machine Learning
Models, agents, and AI applications with specialized lifecycle metadata.
| Type | Description |
|---|---|
| ML Model | Machine learning model (XGBoost, Transformer, CNN, etc.) |
| AI System | Registered AI system with EU AI Act risk classification |
| LLM Agent | Large Language Model agent or chatbot |
| AI App | AI-powered application |
These types unlock the Model Details tab with architecture, framework, deployment status, hyperparameters, and evaluation metrics. Learn more →
⚡ Feature Store
ML feature engineering platforms and feature groups.
| Type | Description |
|---|---|
| Feature Store | Feature serving platform (Feast, Tecton, etc.) |
| Feature Group | Collection of related ML features |
Feature types get specialized tabs for feature groups, materialization tracking, entity definitions, and store sync status. Learn more →
🎬 Media & Content
Rich media datasets for computer vision, NLP, audio processing, and content production.
| Type | Description |
|---|---|
| Video | Video dataset or recording |
| Image | Image dataset or collection |
| Audio | Audio dataset or recording |
| Text Collection | Text corpus or document collection |
| 3D Model | 3D model or point cloud |
Media types unlock the File Metadata tab with type-specific properties (resolutions, codecs, sample rates), folder structure, and annotation metadata. Learn more →
💬 Prompt & GenAI
Managed prompt engineering for generative AI workflows.
| Type | Description |
|---|---|
| Prompt Collection | Managed collection of AI prompts (single or multi-turn) |
Prompt Collections extend the Schema tab with prompt-specific field editing: message sequences, template variables, model compatibility, temperature, and token settings. Learn more →
Feature Availability by Type
Not all tabs and features apply to every product type. The table below summarizes key differences:
| Feature | Data/BI | ML Model | Feature Store | Media | Prompts |
|---|---|---|---|---|---|
| Schema | ✅ | — | ✅ (as "Features") | — | ✅ (with prompts) |
| File Metadata | — | — | — | ✅ | — |
| Model Details | — | ✅ | — | — | — |
| Feature Groups | — | — | ✅ | — | — |
| Data Profiling | ✅ | — | — | — | — |
| Quality Health | ✅ | ✅ | ✅ (as "Feature Health") | ✅ | ✅ |
| Lineage | ✅ | ✅ | ✅ | ✅ | ✅ |
| Governance | ✅ | ✅ | ✅ | ✅ | ✅ |
All types share governance, lineage, quality, versioning, collaboration, and access features.
Custom Product Types
Space administrators can create additional types through Space Settings → Product Types. Custom types support:
- Name — unique within the space
- Description — explains what this type represents
- Icon — chosen from the Lucide icon set
- Color — hex color for visual distinction in the UI
- Default Workflow — auto-routes approval requests for products of this type
Custom types behave like standard data infrastructure types, receiving the universal feature set without specialized extension metadata.
Learn More
- Data Infrastructure Types — Tables, dashboards, APIs, and more
- AI & ML Types — ML Models, AI Systems, LLM Agents
- Feature Store Types — Feature Stores and Feature Groups
- Media & Content Types — Video, Image, Audio, Text, 3D
- Prompt & GenAI Types — Prompt Collections