Google DeepMind’s AI-powered protein design system, AlphaProteo, is taking confident and rapid steps toward making a real, tangible medical impact on the ground. Isomorphic Labs, the Google spin-off specializing in drug discovery, is preparing to begin human clinical trials on drugs designed entirely via artificial intelligence techniques by the end of 2026. This development marks a major milestone in the history of modern medicine, combining the immense computational capabilities of the machine with a deep understanding of complex human biology.
From lab to clinic: the journey of innovation
At the Wired Health event held in London this week, Max Jaderberg, head of Isomorphic Labs, stated that the company’s generalizable drug design engine, known as “IsoDDE,” has achieved “breakthrough performance in the core foundational capabilities of the drug discovery process.” This announcement marks a major turning point for the AlphaProteo system, which was first introduced to the public in September 2024 as an advanced AI system capable of designing new proteins that bind to target molecules with unprecedented precision in the history of medical science.
AlphaProteo achieves binding capabilities that outperform previous best methods and techniques—considered the gold standard in the field—by 3 to 300 times. These superior capabilities have been demonstrated across 7 target proteins rigorously tested in strict laboratory environments. Among the most prominent achievements of this innovative system is the first-of-its-kind success in designing a binder protein targeting VEGF-A, a protein closely linked to cancer development and serious diabetes complications. In vital laboratory experiments conducted on the viral protein BHRF1, 88 percent of the candidate molecules proposed by the system successfully bound to their specified targets with complete success, proving the model’s high efficiency and reliability.
Strategic partnerships with pharmaceutical giants
Isomorphic Labs’ successes have not been limited to theoretical or laboratory research; the company has also secured massive strategic partnerships with leading global pharmaceutical companies such as Eli Lilly and Novartis. These partnerships aim to leverage the revolutionary platform to develop advanced drug pipelines in oncology.
At the World Economic Forum held this past January, Demis Hassabis, the company’s founder and CEO, stated that the company expects its first AI-designed drugs to enter clinical trials by the end of this year, a timeline revised from an earlier target set for late 2025. This move reflects the complexity and sensitivity of transferring technology from computer screens to patient bodies, as the process requires strict adherence to the highest safety and efficacy standards.
How does the revolutionary technology work?
The core technology powering this innovation relies on a pipeline consisting of two integrated steps: the first relies on a generative transformer model trained on massive databases—specifically the Protein Data Bank—alongside more than 100 million predicted protein structures extracted from the renowned AlphaFold system. The second step is an advanced filtering model that predicts which designs will actually succeed in real physical experiments. This innovative approach allows researchers to generate functional binder proteins in just a single screening round, without the need for additional costly and time-consuming optimization processes.
A broader scientific agenda for AI
AlphaProteo is part of a broader drive led by Google DeepMind to integrate artificial intelligence into the core of scientific research. In an article published by Fortune in February, Hassabis and James Manyika, senior vice president at Google, noted that AlphaFold has already been used by more than 3 million researchers across 190 countries worldwide. Since then, the company has significantly expanded its scientific portfolio, adding new tools such as the AlphaGenome system, which aims to predict how genetic mutations in DNA cause diseases, alongside the “AI-assisted world” technology that helps researchers generate new and innovative research hypotheses.
Reflecting deeply on the pace of this astounding progress, Jaderberg wrote in a LinkedIn post: “Back in 2021, the idea of using AI to design drugs was just a hypothesis. A dream. Today, just 5 years later, we see the results. It actually works.”
FAQ
What is the AlphaProteo system?
It is an advanced system relying on machine learning techniques to design new proteins that can bind to target molecules with exceptional precision, opening new horizons in developing effective treatments for intractable diseases.
When will clinical trials begin for drugs designed using this technology?
Isomorphic Labs is preparing to begin human clinical trials on innovative drugs by the end of 2026, following impressive laboratory successes.
What is the system’s success rate in laboratory experiments?
In laboratory tests on a specific viral protein, 88 percent of the molecules designed by the system successfully bound to their specified targets, and it achieved binding capabilities up to 300 times superior compared to previous methods.
