ACD/Labs Percepta pKa predictions have been the industry standard for 25 years. The algorithm is continually improved. In this presentation, Kiril Lanevskij discusses collaborations with pharma and chemical companies that have helped enhance the algorithm. In particular, he shares the inclusion of data in the v2023 release which incorporated 3500 molecules, expanding the training set to 25,000 compounds. Hear about the significant upgrades in speed, accuracy, and reduction of normalization errors in the release, and how this enhanced predictions for properties beyond pKa. See also, evaluation of v2023 with a diverse dataset.