Tableau-Cortex Analyst Alignment — Phased Approach (Definitional Contract → Generated Views → Native Consumption)
Decision
Adopt 4-phase Tableau-Cortex Analyst alignment for Runwal: Phase 0-1 (Definitional Contract — parity test with Semantic View as canonical, Sales first, this quarter), Phase 2 (Generated Reporting Views — RNWL_REPORTING.VW_* regenerated from Semantic View definitions, Q1 FY27), Phase 3 (Governance — release checklist + nightly parity job + KPI register), Phase 4 (Native Tableau semantic view consumption piloted late FY27).
Rationale
Discovery via tableau-mcp + snowflake-mcp confirmed two parallel reporting layers consuming the same upstream Salesforce/SAP staging data independently. Tableau site subodhrunwalgroup queries 28 hand-built SQL views in RNWL_PROD.RNWL_REPORTING; Cortex Agents RUNWAL_BOT (Sales, Nov-2025) and RUNWAL_CRM_BOT (CRM, 06-May-2026) query Semantic Views V2 in RNWL_PROD.RNWL_SALES.RNWL_SALES_REPORTING and RNWL_PROD.RNWL_CRM.RNWL_CRM_REPORTING. No enforcement that measures (e.g. MTD bookings) match across surfaces — drift is structural. Phased approach: validation now → unification next quarter → native long-term. Sequenced to CTO’s FY27 data-foundation focus shift. Construction (RNWL_CONSTRUCTION_REPORTING already exists) starts aligned by design.
Alternatives Rejected
REJECTED: (a) Big-bang consolidation now — too disruptive given 32 active workbooks updated daily; (b) Status-quo with manual periodic checks — drift compounds silently and Cortex Analyst credibility degrades; (c) Wait for Tableau native semantic-view consumption — preview features not production-ready, blocks progress for 6-12 months.
Outcome
Pending