Poultry processing plants utilize batchers to produce products, aiming to minimize operational costs while adhering to a target throughput imposed by their customers. This paper studies the production control of such batchers, leading to an online, time-constrained, and bi-objective decision problem. To solve this problem, we introduce a genetic algorithm equipped with local search and utilize a novel cost prediction function. This hybrid genetic algorithm (HGA) is shown to be competitive with complete enumeration approaches, and outperforms other heuristics in small-scale experiments. We also conduct a numerical analysis that evaluates the sensitivity of costs and throughput to the problem setting. Finally, an industry case study shows that HGA significantly reduces costs, and nearly eliminates deviations from the imposed throughput compared to current practice.