Skip to content

Chat

Cloud Avenue Chat is an AI-powered conversational interface for data analysis and exploration. Ask questions in natural language and receive insights, visualizations, and actionable information from your connected data sources.

Overview

The Chat feature enables you to:

  • Query databases using natural language instead of writing SQL
  • Generate visualizations with charts and graphs
  • Analyze lineage to understand data dependencies
  • Save and share artifacts for future reference
  • Switch between AI models for different use cases

Getting Started

Creating a Chat Session

  1. Navigate to the Chat page from the main navigation
  2. Click "Start New Conversation" or the "+" button
  3. Begin typing your question or request

Each chat session maintains its own conversation history and context, allowing you to have multiple ongoing analyses.

Selecting a Data Source

Before querying data, you need to select which database to use:

  1. Click the database icon in the chat title bar (top center)
  2. Choose a data source from the dropdown menu
  3. The selected source is displayed as a chip below the title bar

Once selected, the AI will query this database for all data-related questions in the current session. You can change the data source at any time by clicking the database icon again.

Demo Data Source

A demo data source ("[Demo] Jaffle Shop") is pre-configured for testing. You can use it to explore features before connecting your own databases.

Data Source Persistence

Your data source selection is saved with the chat session, so it will be remembered when you return to the conversation.

Model Type Selection

Cloud Avenue offers two AI model types optimized for different scenarios:

Light Model

  • Speed: Faster response times
  • Efficiency: Lower computational cost
  • Use cases: Simple queries, quick lookups, straightforward analysis
  • Best for: Exploratory questions, basic reporting

Advanced Model

  • Power: More sophisticated reasoning and analysis
  • Accuracy: Better handling of complex queries
  • Use cases: Multi-step analysis, complex aggregations, nuanced interpretations
  • Best for: In-depth investigations, complex business logic

Switching Models

You can toggle between model types using the switch in the chat title bar:

  1. Locate the "Light / Advanced" toggle in the top-right area
  2. Click to switch between models
  3. The selection applies immediately to new messages

The model type is saved per session, so each conversation can use a different model based on your needs.

Chat Artifacts

Artifacts are structured outputs generated by the AI during your conversation. They can be saved, refreshed, and shared across your organization.

Artifact Types

Cloud Avenue Chat supports three types of artifacts:

1. Lineage Graph

Enterprise Feature

Lineage Graph artifacts are available for Enterprise accounts with connected integrations and GitHub repositories.

Purpose: Visualize data dependencies and relationships across your data ecosystem.

Features:

  • Interactive graph: Pan, zoom, and explore entity relationships
  • Entity types: Tables, views, dashboards, models, and more
  • Dependency tracking: See upstream and downstream dependencies
  • Multi-layer visualization: Understand data flow across systems

Use cases:

  • Impact analysis: What will be affected if I change this table?
  • Root cause analysis: Where does this data originate?
  • Documentation: Visual map of your data architecture

Viewing options:

  • Click an artifact button in the chat to open it in full-screen mode
  • Use graph controls to filter by entity type, search for specific entities
  • Toggle between different view layouts

2. Data Table

Purpose: Display query results in an interactive, sortable table.

Features:

  • Sortable columns: Click column headers to sort ascending/descending
  • Column management: Show/hide columns as needed
  • Search and filter: Find specific rows quickly
  • Row count indicator: See total rows and whether results are truncated
  • Copy functionality: Copy entire table or selected data
  • Export options: Download as CSV or JSON

Generated from:

  • Natural language questions that require data retrieval
  • SQL queries you request
  • Analytical requests that produce tabular results

Modal options (when viewing):

  • Table/SQL toggle: Switch between viewing the data table and the underlying SQL query
  • Dialect indicator: Shows which SQL dialect was used (PostgreSQL, MySQL, etc.)
  • Database name: Displays which database was queried
  • Source name: Shows which configured data source was used
  • Row count: Total number of rows (e.g., "Showing 100 rows")
  • Truncation indicator: Badge showing "Truncated" if results were limited
  • Copy SQL: Copy the SQL query to your clipboard
  • Save artifact: Save this table for future reference
  • Refresh: Re-run the query to get latest data (for saved artifacts)

3. Chart

Purpose: Visualize data patterns and trends through graphical representations.

Supported chart types:

  • Bar charts: Compare values across categories
  • Line charts: Show trends over time or ordered dimensions
  • Pie charts: Display proportions and percentages
  • Scatter plots: Reveal correlations and distributions

Features:

  • Interactive legends: Click to show/hide data series
  • Tooltips: Hover over data points for detailed information
  • Responsive sizing: Charts adapt to screen size
  • Color customization: Change color schemes (via edit controls)
  • Axis formatting: Customize labels, scales, and units

Chart controls (when viewing):

  • Chart type selector: Switch between bar, line, pie, and scatter charts
  • Axis configuration:
  • X-axis: Select dimension or time field
  • Y-axis: Choose measure(s) to plot
  • Series: Group data by category
  • Color customization: Pick colors for data series
  • Number formatting:
  • Decimal places: Control precision (0-4 decimals)
  • Currency: Format as currency with symbol (USD, EUR, GBP, etc.)
  • Percentages: Display as percentage values
  • Legend position: Top, bottom, left, right, or hidden
  • Stacking options: Stack bars or lines for cumulative views

Modal options:

  • Edit controls: Modify chart configuration without regenerating
  • Save artifact: Save this chart for future reference
  • Refresh: Re-run the query and update the chart (for saved artifacts)
  • Export: Download as PNG image
  • View SQL: See the underlying query that generated the data

Saving Artifacts

Artifacts can be saved for future reference:

  1. Click the star icon when viewing an artifact in the modal
  2. The artifact is saved to your account's artifact library
  3. Access saved artifacts from the Artifacts page in the main navigation

Shared Artifacts

Saved artifacts can be viewed by other users in your organization, enabling collaboration and knowledge sharing.

Refreshing Artifacts

For Data Table and Chart artifacts connected to a data source, you can refresh them to get the latest data:

  1. Open a saved artifact from the Artifacts page or from chat
  2. Click the refresh icon in the modal header
  3. The query is re-executed against the database
  4. The artifact updates with current data

Refresh status indicators:

  • Idle: Ready to refresh
  • Running: Query is executing
  • Succeeded: Refresh completed successfully
  • Failed: An error occurred (see error message)

Refresh information displayed:

  • Last refreshed timestamp
  • Who last refreshed the artifact
  • Number of times refreshed
  • Refresh count

Lineage Graphs

Lineage graph artifacts cannot be refreshed as they are snapshots of your metadata at a point in time.

Sample Questions

Cloud Avenue Chat can help with various data analysis tasks. Here are some examples:

Data Exploration

  • "Show me the top 10 customers by revenue"
  • "What's the average order value by month?"
  • "List all tables in the public schema"

Visualization

  • "Create a bar chart of sales by region"
  • "Plot revenue trends over the last 12 months"
  • "Show customer distribution by country as a pie chart"

Analysis

  • "Calculate the year-over-year growth rate"
  • "Find customers who haven't ordered in 90 days"
  • "Compare performance across product categories"

Chat Interface Features

Conversation Management

  • Sessions list: View all your conversations in the left sidebar
  • Search: Find specific sessions by title
  • Delete: Remove sessions you no longer need
  • Auto-save: All messages are automatically saved

Message Actions

  • Retry: If an error occurs, click the retry button on the error message
  • Copy: Copy code blocks or SQL queries
  • Feedback: Provide feedback on AI responses

Title Generation

Chat sessions automatically generate descriptive titles based on your conversation topic. The title appears in the title bar and sessions list.

Artifact Modal

When you click on an artifact in the chat, it opens in a full-screen modal with various controls:

Chart-Specific Controls

  • Chart type selector: Switch visualization type
  • Edit panel (right side):
  • Field mappings (X-axis, Y-axis, Series)
  • Chart options (stacking, colors, legend)
  • Number formatting (decimals, currency, percentage)
  • Export: Download chart as image
  • Resize: Modal is responsive to window size

Best Practices

Effective Prompting

  1. Be specific: Include table names, date ranges, or metrics when known
  2. Provide context: Mention the business question you're trying to answer
  3. Iterate: Refine your question based on initial results
  4. Use examples: Reference existing reports or queries as templates

Managing Sessions

  1. Descriptive titles: Rename sessions to reflect their purpose
  2. Organize by topic: Keep related analyses in the same session
  3. Share insights: Save important artifacts for team access

Data Source Selection

  1. Choose the right source: Select databases that contain relevant data
  2. Test with demo: Use the demo source to test questions before production
  3. Consider performance: Large databases may take longer to query

Working with Artifacts

  1. Save important results: Don't lose valuable insights
  2. Refresh periodically: Keep charts and tables up to date
  3. Document findings: Add context in chat before generating artifacts
  4. Export when needed: Download for presentations or reports

Troubleshooting

"No data source selected" Error

Solution: Click the database icon in the title bar and select a data source before asking data questions.

Queries Taking Too Long

Possible causes:

  • Complex query with multiple joins
  • Large dataset without indexes
  • Database server under heavy load

Solutions:

  • Simplify your question
  • Add filters to reduce data volume
  • Use the Light model for faster processing
  • Contact your database administrator

Incorrect Results

Possible causes:

  • Ambiguous question
  • Missing context
  • Data quality issues

Solutions:

  • Rephrase your question with more specifics
  • Check the SQL query to verify it matches intent
  • Verify data accuracy in the source database
  • Use the Advanced model for complex analysis

Artifact Won't Refresh

Possible causes:

  • Data source no longer accessible
  • Database connection issue
  • Permissions changed

Solutions:

  • Verify data source connection in Settings
  • Test connection from Data Sources page
  • Check with database administrator

Privacy and Security

  • Data isolation: Your chat sessions and artifacts are private to your account
  • Data encryption: All data is encrypted in transit and at rest
  • Data retention: We only store a snapshot of the data needed in an encrypted format for the artifact to function. This data is deleted when the artifact is deleted.

Support

For assistance with Chat features: