Another AI Drug Discovery Partnership: Mankind Puts Early Molecular Design Into a Verifiable Process

Against a backdrop of high drug development costs and high failure rates, pharmaceutical companies’ interest in AI is driven not only by the promise of speed, but also by a more practical question: whether unsuitable designs can be eliminated earlier, before candidate molecules enter costly experimental and clinical processes.

Indian pharmaceutical company Mankind Pharma and Denovo Sciences announced an AI-driven drug discovery collaboration. According to a report by The Economic Times, the two sides will combine Denovo’s AI platform with Mankind’s R&D and validation infrastructure for molecular generation, evaluation, and optimization in the early stages of drug discovery.

The core of this type of collaboration is not to have algorithms “invent” drugs on their own, but to place models within an R&D loop involving human personnel. AI can propose molecular candidates, predict certain properties, or help rank options; research teams then adjust direction based on synthesizability, activity, toxicity risks, and subsequent experimental results. The focus of the human-in-the-loop model described by the two companies is to ensure that computational outputs continuously undergo human scientific judgment and experimental feedback.

However, publicly available information remains quite limited. The report did not disclose the disease areas targeted by the collaboration, the drug targets, the datasets used, model performance metrics, or whether any candidate molecules have entered in vitro or animal validation. Therefore, this collaboration is better understood as the deployment of early-stage R&D capabilities, rather than as proof of concept having been achieved for a specific drug asset.

For Mankind, this arrangement may help embed AI tools into existing R&D workflows, rather than creating a separate technology track disconnected from the laboratory. For Denovo, connecting with a pharmaceutical company’s validation platform is a key threshold that AI drug discovery companies must face: molecules proposed by models ultimately still need to stand up in biological experiments, medicinal chemistry, and manufacturing realities.

**Background Context**

In recent years, AI drug discovery has gradually moved from being a capital markets buzzword into a period of more rigorous scrutiny. The industry is no longer concerned only with how many molecules a model can generate, but whether those molecules can be synthesized, whether they have sufficient selectivity, whether their effects can be reproduced in disease-relevant models, and whether they can translate into interpretable efficacy and safety signals in subsequent clinical development.

This is also the most practical limitation of such collaborations. AI can shorten the time needed for searching and ranking, but it cannot replace wet-lab experiments, clinical design, or regulatory review. If the two sides can disclose specific targets, validation data, and advancement milestones for candidate molecules in the future, the market will have a clearer basis for judging whether this is an efficiency improvement in the R&D process or a deeper change significant enough to alter the success rate of drug discovery.