GenAI-Assisted BI Migration with Guardrails

Tableau → QuickSight Migration Assistant

Built a migration assistant that maps condensed Tableau metadata to QuickSight datasets, calculations, and visuals; generates reviewable deployment packages and ordered API plans; runs five validation dimensions including metric parity fixtures; gates deploy on confidence and structural checks. GenAI behind a swappable interface (rule-based mock default; guarded Bedrock path). deploy.py defaults to dry-run.

★PASS Deploy Gate
Start a project like thisView source on GitHub

Challenge

Tableau → Amazon QuickSight migrations stall on manual remapping of calculations, visuals, and Amazon Q topics—teams ship incomplete assets or break executive metrics without parity checks before production.

Solution

Built a migration assistant that maps condensed Tableau metadata to QuickSight datasets, calculations, and visuals; generates reviewable deployment packages and ordered API plans; runs five validation dimensions including metric parity fixtures; gates deploy on confidence and structural checks. GenAI behind a swappable interface (rule-based mock default; guarded Bedrock path). deploy.py defaults to dry-run.

Impact Metrics

PASS
Deploy Gate
5
Validation
check types
1%
Parity Tolerance
Dry-run
Default Mode

Results

  • • migration_report.json deploy_allowed: true on demo package
  • • 5 validation types — structural, datasource, calculation, parity, visual
  • • 1% tolerance metric parity on high-confidence, non-LOD calcs
  • • Ordered Create* API dry-run plan without AWS mutations
  • • Documented gaps for LOD, table calcs, and pixel-perfect layouts

Business Impact

Accelerates QuickSight adoption with reviewable artifacts and hard deploy gates—so teams migrate with eyes open, not surprise broken KPIs.

Architecture

Tableau → QuickSight Migration AssistantMetadataMappersGenAIPackageValidatorPASS5 checks: Structural • Datasource • Calculation • Parity 1% • Visualdeploy.py — dry-run API plan (default)

Pairs with Tableau Workbook Knowledge Platform for documentation + migration.

Map → Generate → Validate → Deploy

A proven approach combining statistical rigor, automation, and AWS best practices.

1

Metadata Mapping

Datasource, calculation, and visual mappers from Tableau metadata to QuickSight asset YAML.

2

Package Builder

Assemble manifest, layout gap report, and Amazon Q topic definitions for human review.

3

Validation Engine

Structural references, datasource class checks, calc confidence gating, parity CSV fixtures, visual filters.

4

Deploy Planner

Ordered QuickSight API plan; --dry-run default; --deploy-dev guarded for real AWS.

Scale & Scope

Demo
Checks
metadata — end-to-end run_migration.py loop
5
Parity
validation dimensions — FAIL blocks deploy
Explicit
Deploy Mode
non-goals — LOD, table calcs, ATTR() documented

Technology Stack

  • Python mapping + validation engine
  • YAML migration artifacts
  • Amazon QuickSight API planner
  • Guarded Amazon Bedrock generator path
  • Parity fixture CSVs

Need a similar solution?

Let's replicate this success within your organization with a tailored engagement plan.

Start a ConversationView Our Process