One such solution is Ceres TCH, which estimates agricultural productivity by modeling satellite images, Unmanned Aerial Vehicles (UAVs), weather data, and countless other indicators from the field. This information is the key to defining the entire agricultural planning process, together with the operational areas, in order to ensure the predictability of industrial production.

Another AI solution that has been developed supports the identification of planting gaps and invasive plants by processing UAV images with an accuracy of 2.5 centimeters. These images are processed in a cloud environment and indicate the exact location and percentage of gaps and weeds. After this, several adjustments are made in handling the soil and in agricultural processes to ensure these effects are minimized, seeking to maximize our productivity.

Aiming to obtain greater assertiveness in the purchase of by-products and optimize logistical freight costs, we developed a new demand prediction model. Given all possible combinations between city, product, and segment, there are more than 12,000 mathematical models, acting together, to predict fuel demand in the locations where we operate.

To support ethanol and sugar trading, we have also developed advanced analytical solutions such as neural networks that consider thousands of market and internal variables to predict economic indices that allow us to maximize value in this trading. The application of these models has the potential to optimize sales results between 4% and 5%.

In our relationship with resellers, we have developed a number of algorithms that identify the best opportunities to create value in each of the Shell service stations, guiding the commercial agenda to capture this result. It is a win-win relationship with resellers, providing greater assertiveness in the commercial relationship.

For the proximity market, we have also started to develop advanced analytical solutions to measure the effect of promotions at Shell Select stores conducted jointly with their suppliers. Another application being developed is the identification of the risk of stock-outs, allowing stores to anticipate stock-outs and ensure higher sales and a better experience for our customers.