Zygos is an AI tool that lets you rapidly evaluate the likely biological activity of any supplement or pharmaceutical—even when published data are sparse.
This is because it uses an advanced AI model that analyses how a molecule’s atomic arrangement influences its behaviour in the human body.
At its core is a specialised Neural Network designed to handle structured data such as molecules. By treating each molecule as a network of connected atoms, it “learns” the shape and chemical context, enabling accurate predictions of its binding affinities—even for compounds the model has never encountered.
During training, the Neural Network ingests thousands of molecules whose structures and binding affinities are already known. Over many iterations, the model begins to recognize recurring substructures—for example, that certain ring systems paired with electron-donating groups tend to bind strongly to specific receptors—and applies these learned patterns to forecast the behaviour of entirely new molecules.
The result of extensive training and careful hyperparameter tuning is a powerful tool that enables users to quickly evaluate how a molecule is likely to interact with various receptor sites in the body. Furthermore, it may even predict the behaviour of molecules not yet observed in clinical data.
The image below displays Zygos’ predictions for the potent psychedelic 5-MeO-DMT. The model accurately identifies broad serotonergic activity, with a strong emphasis on the 5-HT2 receptors—key components known to play a central role in the mechanism of action of psychedelics.
What does Zygos do?
Zygos accepts input in the form of a molecule represented by SMILES notation (Simplified Molecular Input Line Entry System)—a compact way of describing chemical structures using a string of letters, numbers, and symbols. From this input, Zygos predicts the molecule’s biological behavior, specifically by estimating pKi binding affinities across various biological targets. It also provides a visual depiction of the molecule’s structure.
In simple terms, pKi is a measure of how strongly a molecule can bind to a specific receptor or protein in the body. The higher the pKi value, the stronger the binding. This is especially relevant in the context of the brain, where many medicines and supplements act by targeting receptors that influence mood, perception, alertness, and other cognitive functions.
For example, a high pKi at serotonin receptors—particularly the 5-HT2 family—can suggest that a molecule may have psychedelic, antidepressant, or other mood-altering effects, depending on the receptor subtype involved.
By estimating pKi values, Zygos helps researchers and clinicians better understand how potent a substance may be and which receptors it’s likely to affect—providing valuable insights into its potential effects and side effects in the brain.
How do I use Zygos?
Using Zygos is straightforward. To get started, simply locate the SMILES notation for the molecule you’re interested in and enter it into the designated input box. Press Enter, and the model will begin processing your request. For particularly complex molecules, results may take a few seconds to appear. You can begin making predictions for free using the Basic model.
Finding a molecule’s SMILES notation is usually easy, as it’s a widely used standard in chemistry. For well-known pharmaceuticals and supplements, you can often find the SMILES string in the Wikipedia entry under the Chemical and physical properties section. For less common or more obscure compounds, it’s best to consult PubChem, a comprehensive chemical database. There, the SMILES notation is listed under the “Names and Identifiers” section of a compound’s page.
Zygos for industry stakeholders
Zygos benefits both end users and industry stakeholders in compelling ways. For consumers of pharmaceuticals and supplements, it offers a data-driven way to gauge a compound’s likely effectiveness and safety before investing time or money—helping individuals choose products with stronger scientific backing and potentially avoid ineffective or risky options.
For investors in the pharmaceutical sector, this technology provides an early-warning system and a competitive edge: by rapidly screening large libraries of medicine candidates in silico, firms can prioritize the most promising compounds, de-risk their R&D pipelines, shorten development timelines, and allocate capital more efficiently—ultimately improving the odds of successful clinical outcomes and maximizing return on investment.