Forget the long, painstaking waits that have defined drug and agricultural chemical discovery for decades. Now, thanks to a partnership between Google Cloud and Evogene Ltd, we\u2019re on the brink of an era where powerful artificial intelligence designs new molecules with speed and ingenuity that was unthinkable a few years ago.<\/p>\n
This breakthrough surfaced publicly on June 10, 2025, but in truth, it\u2019s the culmination of years of work. Evogene\u2019s ChemPass AI platform\u2014now running with Google\u2019s backbone\u2014gives researchers something akin to a supercharged laboratory. With it, they can seek out molecules optimized for everything at once: potency, safety, stability, even patentability. All these typically conflicting traits get woven together in the earliest stages, reducing the risk and guesswork that\u2019s historically made drug and agricultural chemical R&D so slow, costly, and frustrating.<\/p>\n
For comparison: in the old days (that is, pre-2025), scientists would test a new compound for effectiveness, only to discover after many more rounds of screening that it wasn\u2019t safe, or didn\u2019t last in the body, or was nearly identical to something on the market. That\u2019s probably why over 90% of drug candidates never made it to pharmacy shelves. Narrow chemical \u2018comfort zones\u2019 and sequential testing stifled true innovation. So the urgent need, across pharmaceuticals and agriculture alike, has always been to escape this bottleneck.<\/p>\n
What really sets this new generation of AI-driven discovery apart is how it rewrites the rules. Evogene\u2019s latest model is based on AI architectures similar to those used for language (the kind that let you chat with GPT or translate text at a click), but it speaks the language of chemistry. It was trained on an immense dataset: around 40 billion molecular structures<\/b>. That\u2019s an ocean of chemistry, and the AI can swim through it with ease, generating entirely new molecules represented as SMILES strings, the shorthand chemists use to describe a molecule in text.<\/p>\n
But it\u2019s not just about numbers. The model is inventive and precise. In tests, about 90% of the AI-created molecules ticked all the right boxes<\/b>\u2014a huge jump from the 29% hit rate reported for older, generic machine learning models. And ChemPass AI doesn\u2019t rely solely on one approach: while its \u201cimagination\u201d comes from models that sequence chemical structures, its reliability is sharpened by graph neural networks, which are perfect for analyzing the connections that make molecules stable, effective, and patent-worthy.<\/p>\nMulti-Trait Optimization: The Quiet Revolution<\/h2>\n
If you ask researchers what really excites them, many will point to ChemPass AI\u2019s ability to balance multiple criteria without laborious back-and-forth. The AI is \u201crewarded\u201d in its training for finding molecules that excel in every criteria\u2014potency, safety, bioavailability, and more\u2014simultaneously. That means instead of picking a compound for just one trait and crossing their fingers, scientists can start with molecules far more likely to survive the long journey to commercial use. Those looking to invent the next blockbuster medicine, eco-friendly pesticide, or advanced material, finally have a tool that thinks as broadly, and as creatively, as they do.<\/p>\n
This work on generative AI for chemistry isn\u2019t isolated. Evogene is also building AI engines for microbial discovery and genetic engineering, forming a Swiss army knife for biotech innovation. These platforms routinely scan billions of genetic and chemical possibilities and score them for their real-world usefulness\u2014something simply impossible for human teams to do on their own, no matter how large.<\/p>\n
In the near future, expect AI-powered digital design to merge even more tightly with laboratory experiments, fueling a virtuous cycle where both machine and human intuition accelerate discovery. It\u2019s not just life sciences that stand to be transformed: as this technology matures, other fields\u2014think food, materials, sustainability\u2014may be next in line.<\/p>\n