Skip to main content
Version: v0.1.0

Chat Assistant

The Chat Assistant is an AI side panel chat accessible from anywhere in D.Hub. It automatically recognizes the context of the current page and allows you to reference datasets/collections/codes/pipelines with @mentions in your questions.

Difference from Knowledge Chat

The Chat Assistant provides AI conversations about the platform in general, while Knowledge Chat provides RAG-based conversations grounded in documents collected in a specific Knowledge. The two features use different backend services.

CategoryChat AssistantKnowledge Chat
Access LocationAll pages (side panel)Knowledge Detail > Chat tab
Backenddhub2-chatwaydhub2-knowledge
Data SourcePlatform metadata (schemas, relationships, etc.)Collected document chunks (RAG)
@MentionsDataset, Collection, Code, PipelineNot supported (use source filters)
VisualizationCharts, maps, graph nodesCitations

Opening the Side Panel

The Chat Assistant is accessible from every page in D.Hub.

  • Click the AI icon on the right side of the screen to open or close the side panel
  • Panel width is automatically adjusted by the layout
  • The panel is displayed alongside the current work screen without interrupting your workflow

Context Awareness

The Chat Assistant automatically detects the page you're currently viewing and provides conversations related to that topic.

PageContextSuggested Question Example
HomeOverall system overview"Summarize the current system status"
OntologyEntities, relationships, graphs"Explain the main entity relationships in this ontology"
CollectionDatasets, access control"Analyze the dataset schemas in this collection"
PipelineWorkflows, transformations, scheduling"Explain the data flow of this pipeline"
OtherGeneral Data Hub features"Tell me what I can do with D.Hub"

When the panel opens, suggested questions tailored to the current context are automatically displayed. Click a suggested question to start a conversation immediately.

AI Suggested Questions

Click the Generate AI Suggestions button to have AI generate customized questions based on the current page context.


Referencing Resources with @Mentions

When typing @ in the conversation input, a mention popup appears to search and select platform resources.

Referenceable Resource Types

TypeIconDescription
CollectionFolderCollection metadata (name, alias, tags, description)
DatasetDatabaseDataset metadata and schema information
CodeCodeCode artifacts (name, language, group)
PipelineNetworkPipeline metadata (name, alias, group)

How to Use

  1. Type @ in the input field to open the mention popup
  2. Type the resource name to search
  3. Click the desired resource or press Enter to select
  4. You can reference multiple resources simultaneously (toggle with Space, confirm with Enter)

Alternatively, click the + button to the left of the input field to open a resource selection popover. This method is convenient for browsing the full list grouped by type.

Selected resources are displayed as tags above the input field, and individual tags can be removed with the X button.

tip

When you @mention a dataset, the AI automatically references that dataset's schema to provide more accurate answers. Example: "@sales_data Write a SQL query to aggregate monthly revenue from this dataset"


Visualization Rendering

The Chat Assistant automatically renders visualizations included in AI responses.

Supported Visualization Types

TypeDescription
Bar ChartValue comparison by category (vertical bar chart)
Line ChartTime series trends (including area graphs)
Pie ChartComposition ratios (pie/donut chart)
Map ViewMap visualization of location data (latitude/longitude markers)
Graph NodeNode reference highlighting in the ontology graph explorer
Ontology Node References

On the ontology page, entities/nodes mentioned by AI during conversations are automatically detected and displayed as a Node Reference Panel below the response. Clicking a node focuses it in the Graph Explorer. This feature only works when the Graph Explorer is open.


Conversation Management

  • Response time is displayed for each conversation to monitor AI performance
  • A stop button allows you to cancel responses being generated
  • Conversation history is maintained while the panel is open

Thinking Indicator

While AI is generating a response, a 4-stage progress status is displayed:

  1. Thinking
  2. Searching Database
  3. Processing Results
  4. Generating Response

Next Steps