Semaglutide: Discovery, Regulatory Approvals, and AI's Transformative Impact on Pharmaceutical Processes

Semaglutide is a long-acting glucagon-like peptide-1 (GLP-1) receptor agonist developed by Novo Nordisk and is an established therapy for type 2 diabetes mellitus (T2D) and obesity. Across the semaglutide unabated sustainability (SUSTAIN) program, subcutaneous semaglutide produced clinically meaningful reductions in glycated hemoglobin (HbA1c), generally in the range of approximately 1.2–1.8 percentage points depending on dose, background therapy, and comparator. In semaglutide treatment effect in people (STEP 1), adults with overweight or obesity without diabetes who received semaglutide 2.4 mg once weekly lost a mean of 15.3 kg (14.9% of baseline body weight) at 68 w. This review summarizes semaglutide's discovery, mechanism of action, pharmacokinetics, clinical trial outcomes, regulatory history, and drug-device strategy, before turning to how artificial intelligence (AI) is now being used to design its successors. We look at generative models such as peptide variational autoencoders (PepVAE) and protein generative pre-trained transformer (ProtGPT2), AlphaFold3-based structure prediction, and multi-objective reinforcement learning from two angles: retrospectively, asking whether these methods would have arrived at semaglutide's own well-characterized design; and prospectively, weighing their potential to cut discovery timelines and cost against published estimates. Taken together, the two halves of this article use semaglutide both as a clinical case study and as a benchmark against which AI-driven drug design can be measured.