We built an iterative scheme of “modeling->screening->validation” through the collaboration within the AD4 consortium. More than 3,000 compounds were physically screened through automated imaging of the 3D Alzheimer’s in a Dish™ assay developed by AD4 consortium collaborators, and nearly 30 compounds have shown an ability to significantly reduce the level of phosphorylated tau, also known as p-tau, in the 3D Alzheimer’s disease (AD) model. Biochemical and toxicity validations not only confirmed some of these compounds’ ability to clear p-tau, but also helped exclude those compounds with significant neurotoxicity and cell death. Interestingly, the validation process revealed the somehow counterintuitive but intriguing fact that more than half of the confirmed hits reduce p-tau induced by amyloid beta without significantly impacting amyloid beta levels in any way, and even more hits do not impact insoluble amyloid beta 42, long considered the main culprit for initiating p-tau accumulation. Such discoveries indicate potential novel mechanism of actions for AD drugs directly targeting tauopathy; this warrants detailed modeling using multiple-omics profiles for drug effects.