Artifacts

Artifacts are visual outputs the assistant creates to help you understand your data. When analyzing logs or investigating issues, the assistant generates charts, tables, and detailed views that appear inline in the conversation.

Artifact types

The assistant can create several types of artifacts:

Timeseries charts

Visualize log counts, error rates, and other metrics over time. Charts support:

• Bar and line modes: Toggle between chart types using the dropdown menu
• Grouped data: View metrics broken down by service, severity, or other fields
• Interactive legends: See total counts and individual series values
• Time range display: View the query parameters used to generate the chart

Example prompts that generate timeseries charts:

"Show me error rates over the last 24 hours"
"Graph request volume by service for the past week"
"Error count by severity, grouped by hour"

Log tables

Display log entries in a sortable, filterable table format:

• Sortable columns: Click column headers to sort by timestamp, severity, or service
• Expandable rows: Click any row to view the full log payload as formatted JSON
• Inline filtering: Filter by severity level or service name using the filter menu
• Active filter chips: See and remove active filters with one click

Example prompts:

"Show me the last 50 error logs from the API service"
"List all WARN and ERROR logs from the past hour"

Log detail

View a single log entry with full context:

• Severity indicator: Color-coded severity badge
• Timestamp and service: Metadata badges at the top
• Full payload: JSON viewer with syntax highlighting

The assistant uses this when drilling into a specific log entry you're investigating.

Diagrams

The assistant can create Mermaid diagrams to visualize flows, sequences, and relationships:

• Multiple diagram types: Flowcharts, sequence diagrams, state diagrams, and Gantt charts
• Pan and zoom controls: Navigate large diagrams with built-in controls
• Fullscreen mode: Expand diagrams for detailed viewing
• Drag to pan: Click and drag to move around the diagram

Example prompts:

"Draw a sequence diagram of the authentication flow"
"Show a flowchart of how requests are routed"

Interacting with artifacts

Charts

Use the chart type dropdown in the header to switch between bar and line views. Hover over data points to see precise values in the tooltip. The legend at the bottom shows total counts.

Tables

Click any row to expand it and see the full log payload. Use the filter button to narrow results by severity or service. Sort columns by clicking the header. The footer shows how many logs match your current filters.

Log detail and JSON viewers

Use the copy button in the top-right corner to copy the formatted content to your clipboard. Line numbers help you reference specific parts of the payload.

Diagrams

• Zoom: Use the + and - buttons in the bottom-left corner
• Pan: Click and drag anywhere on the diagram canvas
• Reset: Double-click or use the reset button to return to the default view
• Fullscreen: Click the expand button in the header for a larger view

Artifact persistence

Artifacts are part of the thread history. When you return to a thread, all artifacts regenerate from the saved query parameters. This means:

• Artifacts reflect the data as it existed when the query ran
• You can revisit past investigations and see the same visualizations
• Shared threads include all artifacts the assistant created

Sharing artifacts

Artifacts are included when you share a thread with teammates. They see the same charts, tables, and diagrams you see, giving them full context for your investigation.

Thread permalinks preserve artifacts, so you can link directly to a conversation that includes visualizations of the issue you found.

Requesting artifacts

You can ask the assistant to create specific artifact types:

"Create a chart of..."
"Show me a table with..."
"Draw a diagram of..."
"Visualize the error trend..."

The assistant automatically chooses the most appropriate artifact type based on your question. For aggregations and trends, it typically creates timeseries charts. For listing individual entries, it creates tables.