Variational AI Secures $5.5 Million to Expand Foundation Model for Small Molecule Drug Discovery
Variational AI has announced the completion of a $5.5 million Seed extension round to support the market expansion of its generative AI foundation model, Enki, designed for small molecule drug discovery.
The funding round, which exceeded the company’s initial target, was led by Nimbus Synergies, with participation from Merck Global Health Innovation Fund, Quimby Investments, Threshold Impact, and Defined Capital. Existing investors, including Flying Fish, A&E Investment, and Nepenthe Capital, also contributed.
Enki is designed to help biopharmaceutical chemistry teams generate and optimize novel molecular structures with reduced reliance on large virtual libraries. The platform operates without requiring initial datasets; users define their Target Product Profile (TPP) by selecting on-targets, off-targets, and desired molecular properties in minutes, after which Enki autonomously generates potential leads. The model is built on an ensemble of generative algorithms trained on decades of experimental data, covering over 570 targets, including GPCRs, kinases, hydrolases, proteases, nuclear hormone receptors (NHRs), and integrins.
See also: The Explosion of Therapeutic Modalities: Small Molecules, Biologics, and Everything in Between
According to the company, Enki employs active learning within the Design-Make-Test-Analyze (DMTA) cycle, using Bayesian optimization to iteratively refine its predictions. This process reportedly enables the identification of potent, selective leads using data from as few as 500 molecules across several learning rounds. Variational AI reports that its partners synthesize and test approximately 20 novel molecules per project within a few weeks, achieving a hit rate of over 50% and a 90% synthetic success rate.
The company emphasizes that while generative AI has transformed fields such as language and image generation, its impact on small molecule drug discovery remains limited. Jason Robertson, Managing Partner at Nimbus Synergies, noted that Variational AI’s model operates with significantly lower computational and data requirements than foundation models in other domains.
The company, based in Vancouver, Canada, was founded by a team with backgrounds in AI research from Google, Microsoft, MIT, Caltech, and D-Wave Quantum.
Topic: AI in Bio
