cxas-configurable-dashboards

CCAI Insights Configurable Dashboards Skill

This skill guides you in authoring, refining, validating, and synchronizing Contact Center AI (CCAI) Insights Configurable Dashboards.

CCAI Insights Configurable Dashboards allow users to build customizable, multi-tab reporting views with rich visualization widgets (Score Cards, Bar/Line charts, Pie charts, Tables, Sankey diagrams) powered by Vega-Lite specifications and SQL queries against conversation metrics.

1. Overview & Declarative YAML Schema

Dashboards are defined declaratively in dashboards.yaml:

version: "1.0"
project_id: "your-gcp-project-id"
location: "us-central1"

dashboards:
  - dashboard_id: "executive_kpis"
    display_name: "Executive Contact Center KPIs"
    description: "High-level summary of inbound call volumes, virtual agent containment, and quality."
    date_range:
      relative:
        quantity: 7
        unit: "DAY"

    root_container:
      display_name: "Root"
      widgets:
        - container:
            display_name: "Overview Tab"
            description: "Operational summary metrics"
            widgets:
              # Tile 1: Total Volume Scorecard
              - chart:
                  display_name: "Total Conversations"
                  chart_visualization_type: "SCORE_CARD"
                  width: 4
                  height: 3
                  data_source:
                    generative_insights:
                      sql_query: "SELECT COUNT(DISTINCT conversation_id) AS total_calls FROM conversations"
                      chart_spec:
                        mark: "text"
                        encoding:
                          text: {field: "total_calls", type: "quantitative"}

              # Tile 2: Top Contact Drivers Bar Chart
              - chart:
                  display_name: "Top Contact Drivers"
                  chart_visualization_type: "BAR"
                  width: 8
                  height: 6
                  data_source:
                    generative_insights:
                      sql_query: >-
                        SELECT issue_category, COUNT(1) AS volume
                        FROM conversations
                        WHERE issue_category IS NOT NULL
                        GROUP BY 1
                        ORDER BY volume DESC
                        LIMIT 10
                      chart_spec:
                        mark: "bar"
                        encoding:
                          x: {field: "volume", type: "quantitative", title: "Calls"}
                          y: {field: "issue_category", type: "nominal", sort: "-x", title: "Category"}