Core concepts

In D.Hub, you organize assets in collections and process and connect data with pipelines and ontologies. Dashboards and AI features use the data prepared this way.

See the Glossary for unfamiliar terms.

Component map

The components relate to each other as follows.

Collections and assets

A collection groups D.Hub assets by project, department, or topic. A collection contains six asset types.

Asset	Role
Dataset	Stores structured data as a table and manages change history as versions.
Code	Stores Python scripts or SQL queries used by pipelines.
Pipeline	Connects and runs data-reading, transformation, and storage steps.
Knowledge	Collects documents used by search and AI Chat.
Dashboard	Displays query results in chart, table, and map widgets.
Agent	Connects tools and actors to an LLM to process requests.

Place assets in collection folders and find them as items in the tree. A role granted on a collection is inherited by its child assets. Each collection is associated with one ontology scope, so entities and relationships are managed within the same boundary.

Data processing

Create a dataset by uploading a CSV, JSON, or Parquet file, or select + Add Item ▾ → Dataset and define a schema. You can also create one from a REST API.

Reuse Python and SQL code by referencing it from multiple pipelines. Ask D.Hub Assistant in natural language to create a code asset, or use AI code generation and AI editing in the code editor. Review generated or edited code before applying or running it.

A pipeline builds a data flow from nodes and edges. A node is a processing step such as reading, transforming, or storing data. An edge connects the flow between nodes. Run a pipeline directly, on a schedule, or in response to an event.

Ontology

An ontology defines the business meaning of data with entities and relationships. An entity represents an object you manage, such as a customer, product, or sensor. A relationship represents a connection between entities. Properties store details about an entity or relationship.

In Ontology Builder, connect entities and relationships to datasets, then query connected paths in Graph Explorer. For example, after defining an Order relationship between Customer and Product, you can see the products ordered by a customer in the graph.

Analytics and AI tasks

A dashboard connects a dataset with SQL or Simple Mode and displays results in widgets. Configure a date range and automatic refresh interval to rerun queries on that schedule.

Knowledge collects web pages, files, and manually written documents. Added documents are processed into searchable chunks. AI Chat finds chunks related to a question and uses them as context for the answer.

An agent connects tools and actors to an LLM to process requests. Deploy an AI agent and run it in its dedicated chat. A workflow agent runs from an event and can request human approval (HITL) according to an actor's confirmation policy.

Related pages

Task	Page
Manage collections and assets.	Collections overview
Create a dataset.	Create and manage datasets
Create Python or SQL code.	Create and manage code
Build a data-processing flow.	Use the Pipeline editor
Define entities and relationships.	Ontology overview
Connect documents to search and AI Chat.	Knowledge Management overview
Visualize query results.	Dashboards overview
Connect an LLM and tools.	Agents overview

Next steps

Choose the path that matches your goal.

• Quick start — Upload sample data and review the result in a dashboard.
• Role-based paths — Choose a learning path for your responsibilities.
• Glossary — Look up terms used in the documentation.