Generate AI-powered banking reports from natural language queries
When the user requests a new report, this skill executes a 7-step pipeline:
Parse the natural language query into structured parameters:
Query ClickHouse Cloud data warehouse:
nl5th0k8zt.germanywestcentral.azure.clickhouse.clouddwhfact_loans, , , , fact_depositsfact_transactionsdim_customersdim_branchesGenerate a structured report with:
For each section with a chart type:
<blackbox tag="chart"> placeholders<chart> elementsValidate against banking data governance rules:
Auto-approve if compliance score >= 0.8. Otherwise request revision.
dwh.report_tracking)sb5.report.events)| Domain | Table | Key Metrics |
|---|---|---|
| loans | dwh.fact_loans | total_disbursed, outstanding_balance, npl_ratio, avg_interest_rate |
| deposits | dwh.fact_deposits | total_deposits, avg_balance, growth_rate, cost_of_funds |
| transactions | dwh.fact_transactions | volume, total_amount, avg_amount, fee_revenue |
| customers | dwh.dim_customers | total_customers, new_customers, churn_rate, avg_lifetime_value |
| branches | dwh.dim_branches | profit, cost_income_ratio, staff_count, efficiency_score |
The pipeline produces: