Our Technology

A foundation model for drug discovery

Built from the ground up by our team and battle-tested against the reality of the lab—our AI is ready to disrupt preclinical drug discovery across all therapeutic areas.

Our deep-learning model simultaneously predicts therapeutic effects, off-target effects and pharmacological properties.

To achieve this, our team assembled and expertly curated a vast database of biological experiments featuring 750 million data points for 2.5 million molecules, and designed a deep neural network able to distill this data into a superhuman understanding of how a molecule's structure can affect an organism, enabling it to predict protein affinity, pathways activation/inhibition, or other complex biochemical interactions, for any new molecule.

This one of a kind foundation model for drug property predictions enables us to rapidly and confidently sift through vast chemical spaces to identify high quality compounds with dramatically higher chances of success in progressing through preclinical drug discovery.

• Proven: Tested by independent laboratories
• Reliable: Unprecedented accuracy, on par with experiments
• Fast: Finds the rare gems in vast chemical spaces
• Versatile: Applicable in any therapeutic area

Services & Collaborations

We offer both service and milestone-based collaboration models, leveraging our confidence in predictions to significantly expedite drug development projects.

We are also looking for private or academic partners to test in vitro compounds that we first discover in silico.

Collaborating labs have confirmed the potential of compounds that we predicted purely in silico, including:

• Anti-SARS-Cov-2 repurposed drugs
• Novel inhibitors of HIV1
• Cancer growth inhibitors

Our system is applicable to any therapeutic area.

Our latest news: out of a virtual screening of diverse libraries of 23M compounds to find NRF2 agonists, 2 out of 2 compounds tested in cells by Aarhus University were confirmed active!

• Virtual High Throughput Screening (HTS)

  Efficiently identify promising candidates from vast compound libraries using our AI-driven screening technology, significantly reducing the resources and time typically required for early-stage drug discovery.

• Virtual ADMET Profiling

  Our models predict most ADMET properties, ensuring that progressing compounds are both effective and safe, thereby streamlining their path through the development pipeline.

• Virtual Off-Target Screening

  By predicting potential off-target interactions, we enhance the safety profiles of therapeutic candidates, minimizing the risk of early failure in the preclinical stage and adverse effects in clinical trials.

• Virtual Hit Analogs and De Novo Screening

  Our AI-driven genetic algorithm swiftly generates and optimize hit analogs as well as novel compounds, facilitating rapid advancements from hit discovery to lead optimization.

• Integrated Service

  By combining hit discovery with lead optimization in a single powerful step, Cortex achieves a leap in cost-effectiveness. This integrated approach dramatically shortens the drug development timeline, enhancing productivity and efficiency.

Contact Us

Longevity

Drug Pipeline for Life Extension and Age-Related Disorders

In our globally aging society, longevity research is the new frontier for medicine. Lifespan enhancing compounds will prevent, delay and reduce the frequency of cardiovascular disease, cancer, and Alzheimer’s disease. These illnesses are the largest burden on human health, and the leading causes of death.

We are using our new technologies and unique expertise in the mechanisms of aging to develop drug candidates for known and novel longevity targets.

We initiated collaborations with leading laboratories in the following fields:

• Oxidative stress
• Autophagy
• Senolytics
• DNA repair
• Age related disorders

Latest development: Cortex unveils the first AI-discovered compound proven to extend lifespan in a whole organism.

Research & Development

AI-Assisted Molecular Docking

In complement to our current AI system, we are building a new technology leveraging the most recent advancements in the field of deep learning to make accurate predictions of ligand-protein binding pose and affinity.

Our new generation of deep neural networks coupled with quantum molecular simulations will allow for blazing fast exploration of new targets, even when no activity data is available, while achieving unprecedented prediction accuracy.

Our prototype already handles detailed force fields and flexible targets—1000x faster than available open source and commercial alternatives.

Retina — our in-house visualisation tool