Background. Peptic ulcer disease still accounted for roughly 8.09 million prevalent cases and 6.03 million disability-adjusted life years worldwide in 2019, and licorice root remains one of the most widely used traditional treatments for it. Two questions stand between that traditional use and a rational dosage form. Which constituents of the root plausibly engage the molecular machinery of gastric injury, and what release rate would a gastroretentive carrier actually have to deliver. Objective. To rank the documented root constituents of Glycyrrhiza glabra against two structurally characterised gastric targets, and to determine the input-rate window over which a floating gastroretentive microsphere would prolong exposure rather than waste it. Methods. Thirteen root constituents and four reference drugs were profiled for calculated physicochemical properties and drug-likeness in RDKit, then docked with AutoDock Vina 1.2.7 into the gastric H+/K+-ATPase (PDB 5YLU, 2.80 A) and human cyclooxygenase-2 (PDB 5KIR, 2.70 A). Each protocol was validated by redocking its co-crystallised ligand. Exposure was then simulated in a one-compartment oral model built from published non-compartmental values for 18beta-glycyrrhetinic acid, with the apparent absorption rate constant treated as the formulation design variable. Results. Redocking reproduced both crystallographic poses, at 0.553 A for vonoprazan in the proton pump and 0.482 A for rofecoxib in cyclooxygenase-2. Eleven root constituents scored between -8.03 and -9.99 kcal/mol at the vonoprazan site, with liquiritin, glabrene and glabridin ranking above vonoprazan itself at -9.25 kcal/mol, while glycyrrhizin returned a positive score and a truncated pose list at both targets, indicating that neither site could accommodate it. Carbenoxolone, the licorice-derived drug licensed for gastric ulcer and not an antisecretory agent, ranked 5.2 kcal/mol behind vonoprazan at the pump, which is the direction its known mechanism requires. Simulated total exposure was invariant across the input-rate sweep at 15.14 mg.h/L, so slowing input redistributed exposure rather than increasing it. Time above the internal reference concentration rose from 5.26 h at the immediate-release rate to a maximum of 7.06 h at an absorption rate constant of 0.20 per hour, then collapsed to zero below 0.10 per hour as the profile flattened beneath the threshold. At a matched daily dose, twelve-hourly gastroretentive input reduced simulated peak-to-trough fluctuation from 193.4 to 85.3 per cent without altering average steady-state concentration. Conclusion. The gastroretentive design space for this extract is bounded on both sides. An optimum exists near an input half-life of three to four hours, and a carrier that releases more slowly than that trades peak concentration for duration it cannot use. All values reported here are computational predictions and require experimental confirmation.