Repsol

Solution

Repsol built two AI-driven workflows for service-station store operations. The internal Suggested ordering application uses predictive AI and optimization algorithms to estimate replenishment needs for roughly 1,000 items per station and generate order recommendations two or three times a week. About 900 forecasting models and another 900 optimization models run daily, and the system improves as it absorbs new data and learns from decisions made at each station. Recommendations are delivered to station managers, who can accept or adjust them using local context. In parallel, Repsol is piloting Automatic delivery note reading: employees photograph supplier delivery notes on a mobile phone, generative AI extracts fields such as date, supplier, items, and quantities, and the structured output is uploaded into Repsol's system for operator validation. Together, the workflows combine AI automation with human oversight to improve availability and reduce repetitive administrative work.

Data flow

Operational station data feeds daily forecasting and optimization models that produce replenishment suggestions for managers, who accept or adjust the recommendations before orders are placed. For delivery notes, an employee photographs the document on a mobile phone, generative AI extracts date, supplier, items, and quantities, and the validated structured output is uploaded into Repsol's system.