Peptide drugs have revolutionized modern therapeutics, offering novel treatment avenues for various diseases. Nevertheless, low design efficacy, time consumption, and high cost still hinder peptide drug design and discovery. Here, an efficient approach that integrates deep learning-based protein design with functional screening is presented, enabling the rapid design of biotechnologically important peptides with improved stability and efficacy. 10,000 de novo glucagon-like peptide-1 receptor agonists (GLP-1RAs) are designed, 60 of these satisfied the stability, efficacy, and diversity criteria in the virtual functional screening. In vitro validations reveal a 52% success rate, and in vivo experiments demonstrate that two lead GLP-1RAs (D13 and D41) exhibit extended half-lives, approximately three times longer than that of Semaglutide. In diabetic mouse models, candidate D13 results in significantly lower blood glucose levels than Semaglutide. In the obesity mouse model, D13 induces weight loss efficacy comparable to that of Semaglutide. The AI-driven peptide design pipeline-which integrates protein design, functional screening, and experimental validation-reduces the number of iterations required to find novel peptide candidates. The entire process, from design to screening, can be completed in a single cycle within two weeks.