Tableau VDS Driven Self‑Healing Supply Chain Agent

Inspiration

Supply chains drift quickly, but ops teams usually find out too late. We wanted a lightweight, always‑on monitor that connects directly to Tableau Cloud data and auto‑summarizes what to do next.

What it does

• Pulls the latest OTD (on‑time delivery) over a configurable time window
  • Detects threshold breaches
  • Drills down by Shipping Mode + Region and surfaces city/state hotspots
  • Generates a concise investigation summary, recommendations, and an email draft ## How we built it
• Next.js (TypeScript, App Router) UI + API routes
  • Tableau REST API for PAT sign‑in
  • Tableau VizQL Data Service (VDS) for metadata + queries
  • OpenAI Responses API for narrative generation (optional; fallback templates if missing) ## Challenges we ran into
  • VDS schema nuances (filters must use field object)
  • Data field captions differ after publishing → solved with metadata‑based field resolver
  • Recent data gaps → implemented auto‑expanding time window (7→30→90→all‑time)

Accomplishments that we're proud of

• End‑to‑end working MVP in a day
• Stable VDS integration with automatic field matching
• Clear drilldown narrative + actions ready for ops teams

What we learned

• How to structure VDS queries for KPI + drilldown flows
  • Practical PAT auth + metadata matching patterns
  • Building a “self‑healing” workflow with minimal dependencies ## What's next for Tableau VDS Driven Self‑Healing Supply Chain Agent
  • Slack integration + scheduled runs
  • Trend sparkline and alert history
  • More metrics and multi‑datasource support

Built With

• next.js
• openai
• tableau-cloud
• tableau-rest-api
• typescript
• vizql-data-service