Conventional feeding of lactating sows in commercial farrowing units was often constrained by coarse feed control, fixed meal timing, and feed losses caused by rooting and manipulation at the trough. These limitations frequently resulted in unstable feed intake during mid- to late lactation, which in turn increased the likelihood of excessive body-reserve mobilization and delayed reproductive recovery after weaning. To address these problems, this study adopted a two-stage progressive experimental design to systematically evaluate the effects of an intelligent precision feeding system on sow feed intake, body condition, reproductive efficiency, and piglet growth performance, and to identify key operating parameters. The study was conducted in a commercial farrowing facility in Shaanxi, China, from March to June 2025. The intelligent feeding unit integrated controlled-dose dispensing with event-based data logging and a trigger mechanism that delivered small top-up portions when trough residual feed and short-window interaction signals indicated persistent feeding motivation; a water–feed mixing option was implemented through a water-control module. Daily feed intake and daily water consumption were recorded automatically. Sow body condition was assessed using backfat thickness measured at entry to the farrowing crate and at weaning. Reproductive recovery and piglet performance traits were obtained from the farm production-record system. In Experiment I, ninety-four multiparous sows were allocated to an Intelligent Feeding Group (IFG) or a Traditional Feeding Group (TFG) (47 sows per unit) housed in environmentally comparable units. IFG sows were managed using the hybrid logic combining a stage-wise baseline allowance with sow-initiated triggered top-ups, whereas TFG sows were fed with a conventional dry-feed feeder at fixed times (three meals per day). The intelligent system produced more stable intake trajectories, and the clearest separation between treatments emerged during mid- to late lactation. From approximately day 15 postpartum onward, daily feed intake in IFG exceeded that of TFG by about 5%-10%, reflecting a more sustained intake plateau and a later decline toward the end of lactation. Consistent with improved intake stability, body-reserve mobilization was reduced in IFG as indicated by backfat change: mean backfat loss decreased from 2.61 mm in TFG to 1.69 mm in IFG (P<0.01), corresponding to a 35.2% reduction. Reproductive recovery was accelerated in IFG. The weaning-to-estrus interval (WEI) shortened from 9.04 days (TFG) to 8.15 days (IFG) (P<0.01), and the wean-to-service rate increased from 85% to 90%. Piglet outcomes also improved in association with the stabilized maternal intake pattern: average weaning weight increased from 7.30 kg (TFG) to 7.68 kg (IFG) (P<0.01), pre-weaning weight gain rose from 6.06 to 6.47 kg (P<0.05), and pre-weaning survival increased from 89.13% to 92.06%. Together, these results indicated that the intelligent feeding approach supported higher and more persistent feed intake during late lactation and aligned with improved body-condition preservation and superior reproductive and litter performance. In Experiment II, parameter screening was performed within two units equipped with the intelligent feeding system to compare two deployable “meal frequency × dilution” strategies under identical hybrid logic. A five-meal strategy with a water-to-feed ratio of 1.3:1 during mid- to late lactation was compared with a four-meal strategy with a water-to-feed ratio of 1.5:1. The five-meal strategy with a water-to-feed ratio of 1.3:1 maintained a higher and more persistent intake plateau between days 12 and 21 postpartum and delayed the late-lactation decline compared with the four-meal strategy with a water-to-feed ratio of 1.5:1. Daily water consumption showed a similar temporal pattern, with a sharper late-lactation decline under the four-meal strategy with a water-to-feed ratio of 1.5:1. Backfat-change outcomes during parameter screening were consistent with these temporal intake differences. The five-meal strategy with a water-to-feed ratio of 1.3:1 showed a more favorable backfat-loss profile, with smaller loss and a more concentrated distribution with fewer extreme negative values. Because meal frequency and water-to-feed ratio were coupled in this screening comparison, mechanistic interpretation was limited to the combined strategy rather than isolated main effects. In conclusion, a sensor-based, trigger-activated phased precision feeding approach provides a practical and traceable framework for stabilizing feed and water intake during mid- to late lactation in commercial farrowing systems. This approach reduces body-reserve mobilization, as indicated by lower backfat loss, supports faster post-weaning reproductive recovery, and improves piglet growth and survival under the tested conditions. Within the evaluated operating settings, the five-meal strategy with a water-to-feed ratio of 1.3:1 represents a promising deployable configuration for sustaining the late-lactation intake plateau and mitigating the end-of-lactation decline in intensive swine production.