← Back to Home

AbbVie Teams Up with Iambic to Evaluate Small-Molecule Drug Designs Simultaneously Using AI

The companies will use Enchant v3 in drug discovery across immunology, neuroscience and oncology, seeking to identify trade-offs among candidate molecules earlier. The collaboration has not disclosed specific targets or presented validation results showing improved clinical success rates.

By SURL BioNews

Finding new drugs is often challenging because a molecule must meet several requirements at once: it needs the desired biological activity as well as suitable metabolic and safety profiles. AbbVie and Iambic announced a multiyear collaboration on September 21, 2026, aiming to use artificial intelligence to evaluate multiple design hypotheses simultaneously and help research teams decide which small molecules warrant the next round of experiments.

FirstWord Pharma reported the news under a headline about AbbVie adopting Iambic’s new model. According to the companies’ announcements, the collaboration spans immunology, neuroscience and oncology, with the goal of discovering therapies with first-in-class or best-in-class potential. These are research and development directions, however; the announcements did not disclose specific diseases, molecular targets or drug candidates.

The collaboration will use Enchant v3, which Iambic launched on the same day. According to Iambic, the multimodal model has 41 billion parameters, covers more than 6,000 molecular properties and integrates 16 types of biomedical data, including experimental assays, molecular and protein structures, images and clinical trials. It is intended to identify relationships across different data sources to help predict multiple aspects of candidate molecules’ performance; the breadth of properties covered does not, in itself, mean that its predictions have been adequately validated.

In the research and development workflow described by Iambic, the model first makes predictions and assesses uncertainty, followed by chemical synthesis and biological experiments, with the results fed back into the model. This gives AI a role in concrete experimental choices: which compounds are worth making, which unknown properties should be measured first and how to adjust designs to balance multiple requirements.

The validation evidence still has limits. In a technical overview released at the same time, Iambic used its own CDK2/4 and KIF18A inhibitor programs to illustrate how the model contributes to molecular optimization, but these examples are not results of the AbbVie collaboration. The collaboration announcement also provided no prospective comparative data demonstrating how much time Enchant v3 could save or how much it could improve clinical success rates compared with existing workflows. Whether laboratory predictions can translate into benefits for patients remains a question for subsequent research.

The deal structure links the platform’s value to subsequent outcomes: Iambic will receive an upfront payment and will be eligible for milestone payments based on outcomes, as well as tiered royalties on net product sales. Neither the amounts nor the royalty rates were disclosed. The collaboration adds an application for the AI design workflow at a major pharmaceutical company, but its effectiveness still depends on whether it can produce experimentally confirmed molecules that can advance further in development.

References

  1. FirstWord Pharma
  2. AbbVie