Daily Route Delivery Predictor
Built from real-world frustration: a workflow that predicts delivery truck arrivals so a merchandiser never wastes a morning waiting again.
The problem
As a retail merchandiser for the largest wine and spirits distributor in America, I lose hours every week sitting in stores waiting on shipments to arrive before I can stock shelves and build displays. The 7:00 AM delivery projections we receive are rarely accurate.
The solution
I built a workflow that analyzes the past month of delivery data to forecast realistic arrival windows per store. I upload my daily route, it cross-references the historical model, and returns a sharper ETA than the morning projection — so I can sequence my day around when trucks actually show up.
Benefits
- Less downtime, more productive stops per day
- Forecast accuracy improves as more delivery data is captured
- Claude Skills handles the data ingest and automated bug-checks the output for me
Challenges & what I'd improve
I didn't have access to the company's real delivery dataset, so I extrapolated from what I personally tracked over a few weeks to construct a month of representative hypothetical data. With production data access, the model would sharpen significantly.