You're offline - Playing from downloaded podcasts
Back to All Episodes
Podcast Episode

Isomorphic Labs to Begin Human Trials of AI-Designed Drugs by End of 2026

April 27, 2026

0:00
2:28
Podcast Thumbnail

Google DeepMind's drug discovery spinoff Isomorphic Labs says its AI-powered protein design platform is ready for human clinical trials by the end of 2026. The company's IsoDDE engine has achieved binding affinities 3 to 300 times better than previous methods, with partnerships secured with Eli Lilly and Novartis for oncology pipelines.

AI Moves From Lab to Clinic

Google DeepMind's AI-powered protein design platform, AlphaProteo, is moving closer to real-world medical impact as its drug discovery spinoff, Isomorphic Labs, prepares to begin human clinical trials of AI-designed drugs by the end of 2026. Isomorphic Labs president Max Jaderberg announced at WIRED Health in London that the company's generalisable drug design engine, known as IsoDDE, has achieved breakthrough performance on the foundational capabilities essential to drug discovery.

Breakthrough Performance Metrics

AlphaProteo, first introduced in September 2024, is an AI system capable of designing novel proteins that bind to target molecules with unprecedented accuracy. The platform achieves binding affinities 3 to 300 times better than previous state-of-the-art methods across seven target proteins tested in laboratory settings. Among its most notable achievements is the first successful AI-designed protein binder for VEGF-A, a protein linked to cancer and diabetic complications. In wet-lab testing on viral protein BHRF1, 88% of the system's candidate molecules successfully bound to their targets.

Pharmaceutical Partnerships

Isomorphic Labs has secured partnerships with Eli Lilly and Novartis to leverage the platform for oncology pipelines. Founder and CEO Demis Hassabis said at the World Economic Forum in January that the company expects its first AI-designed drugs to enter clinical trials by the end of this year, a timeline pushed back from an earlier target of late 2025.

How the Technology Works

The underlying technology relies on a two-step pipeline: a generative transformer model trained on the Protein Data Bank and over 100 million predicted structures from AlphaFold, followed by a filtering model that predicts which designs will succeed in physical experiments. This approach allows researchers to generate functional protein binders in a single screening round without further optimisation.

A Broader AI Science Agenda

AlphaProteo is part of a wider push by Google DeepMind into AI-driven science. AlphaFold has been used by more than 3 million researchers across 190 countries. The company has since expanded its portfolio with tools like AlphaGenome for predicting how DNA mutations drive disease and AI co-scientist for generating research hypotheses. Jaderberg reflected that what was a hypothesis in 2021 is now producing real results just five years later.

Published April 27, 2026 at 12:15pm

More Recent Episodes