A well funded pizza delivery startup in Mountain View, CA has established a formula for partnering people with technology to create a more thoughtful and efficient way to source, make, and deliver pizza. The pizza is made using only robots and is delivered via food trucks located throughout Silicon Valley.
Needed a way to predict product sales for each truck with the goal of never selling out while minimizing waste.
Location of trucks to maximize sales and reduce delivery times/costs
Timely procurement of ingredients
Plan for territory expansion (beyond Silicon Valley) with possible product diversification.
Forecasted demand for new products to both existing and new territories
Fusemachines to built a system that could predict demand at the ingredient level relative to each market location to determine which meals users would be most likely to buy and where the most effective place would be to have their trucks.
Correlated the effects of weather, location, ingredient costs, days of the week, and holidays had on customers spending. Included marketing spend and conversation analytics into models to help improve demand predictions.
The system reduced the burden on their procurement team as they had better insight into which ingredients and quantity needed.
The system was able to reduce waste across all perishable items by more than 40%, while improving prediction demand at a rate 240% more accurate then previous models. Fusemachines built machine learning models using advanced machine learning and Deep Neural Networks.