In sows’ conventional feeding (CF), diets composition is usually based on the average herd’s nutrient requirements. Thus, sows can be under- or over- fed leading to extra feed costs and environmental losses. Nutritional models and new technology (sensors, automatons), bring opportunities to measure and integrate the individual variability into nutrient requirements estimations. The objective is therefore to go towards precision feeding (PF), combining on-farm data as input for a dynamic nutritional model with smart feeders to provide individual and daily-adjusted rations. A mechanistic model (InraPorc) was upgraded for gestating sows and applied to databases to calculate individual daily nutrient requirements. Herd historical data as well as the animal parity, body weight, backfat thickness and age at insemination were needed to predict some parameters required by the model (i.e. litter size and weight, sow’s target body weight at the end of the gestation). There was a strong inter-and intra-individual variability of the nutrient requirements according to sows’ characteristics, performance and day of gestation. Simulations showed that more sows had their requirements met with PF based on lysine supply than CF, especially for primiparous. With PF, protein intake, feed cost, nitrogen and phosphorus excretions were reduced by 25, 5, 17, and 15%, respectively. The results obtained during an on-farm trial confirmed those obtained by simulation. To improve the accuracy of the nutritional requirements estimations, new parameters could be added to the model like the individual physical daily activity. Indeed, the daily activity varies between and within sows, and impacts the energy requirements. An algorithm is therefore being developed to identify individual activities via video recordings. The results are promising as the neural networks are able to detect a sow lying, standing and eating with accuracies of 82, 73, and 87%, respectively, at the group level. The next step will be to be able to track and identify each individual to obtain its daily activity automatically. Until now, PF concerned only energy and protein supplies. For the future, other nutrient such as minerals and fiber could also be considered, but this requires improvement in the smart feeder design.