Picadeli, a part of Greenfood, is the market leader in Northern Europe with its self-serve salad bars. Currently, Picadeli's salad bars are available across 2000 stores. In each store, the staff estimates the expected demand for salad for the upcoming week.

Estimating product demand poses a challenge for store staff. Overstocking results in unnecessary waste, while understocking risks missed sales opportunities.

The demand for salad varies due to a range of internal and external factors that could not be integrated into Picadeli's previous statistical forecasting models.

Violet's AI solution considers sales data from multiple stores and learns from common patterns such as weekly variations, weather impact, and holidays. The models provide daily forecasts for each store for the next 14 days.

The models are based on Temporal Fusion Transformer and are combined with traditional statistical models. The solution is integrated into the operations and is used in over 1000 stores across Sweden, the USA, Finland, France, and Germany.

Applying AI to Picadeli's forecasting models allows the forecasts to incorporate multiple factors, resulting in higher precision.

A better forecast means less burden on store staff, reduced food waste, and increased salad sales. This, in turn, leads to happier store staff, reduced environmental impact, and improved profitability.