AI-Designed Drugs Move Closer to Human Testing at Isomorphic Labs

A major AI drug-discovery claim reaches a new stage

Isomorphic Labs, the Google DeepMind spinoff built around AlphaFold-driven drug discovery, says it is preparing to start human trials of medicines designed with its AI technology. The update came from company president Max Jaderberg at WIRED Health in London, where he said the startup has built a “broad and exciting pipeline of new medicines” and is now gearing up to enter the clinic.

The statement matters because it moves the AI-drug story from laboratory promise toward clinical accountability. For several years, companies across the biotech sector have argued that machine learning can speed the discovery of better medicines. Human testing is where that thesis becomes measurable.

What Isomorphic is building on

Isomorphic Labs was founded in 2021 as a spinoff from Google DeepMind. Its work is closely tied to AlphaFold, the AI system that transformed protein structure prediction. In 2020, DeepMind presented AlphaFold 2, and the following year an open-source version was released for broad scientific use. In 2024, DeepMind and Isomorphic Labs introduced AlphaFold 3, extending the system beyond proteins in isolation to include molecules such as DNA and RNA and their interactions with proteins.

That progression is central to drug discovery. According to the source text, the platform can help predict how a small molecule may bind to a target and what else it might bind to. Those are core questions in designing medicines that are both effective and safe.

The scientific backdrop is substantial. The source says AlphaFold has predicted the structure of virtually all 200 million proteins known to researchers and has been used by more than 2 million people across 190 countries. Its impact was recognized at the highest level when Demis Hassabis and John Jumper received the Nobel Prize for chemistry.

Why clinical trials are the real test

Despite the excitement around AI in drug discovery, much of the field’s public narrative has rested on promises rather than patient data. Companies can show computational advances, target-selection improvements, and preclinical pipelines, but none of that proves that AI-designed molecules will succeed in humans. That is why Isomorphic’s update stands out.

Jaderberg did not provide a specific timeline during the event, and the article notes that the move comes later than earlier expectations. Last year, CEO Demis Hassabis had said the company would have AI-designed drugs in clinical trials by the end of 2025. Even so, the current message is clear: the company says it is now approaching that threshold.

This is the point where broad claims about efficiency, accuracy, and molecular insight must face the realities of medicine development. Human trials test not just whether a molecule can hit a target, but whether it is tolerated, whether it behaves as expected in the body, and whether the predicted biology translates into clinical benefit.

The broader significance for biotech

If Isomorphic does enter clinical testing soon, the milestone will resonate far beyond one startup. It will offer one of the clearest early signals of whether AI can move from an enabling research tool into a direct engine of therapeutic design.

That does not mean success or failure in a single program will settle the issue. Drug development is too complex for that. But the first wave of clinical evidence will shape how investors, pharmaceutical partners, and regulators evaluate the increasingly crowded field of AI-first biotech companies.

The company’s positioning also matters because it sits at the intersection of elite AI research and drug-development ambition. AlphaFold already changed how researchers study biology. The next question is whether that understanding can be turned into approved therapies at scale.

From protein prediction to medicine

The core promise behind Isomorphic’s work is not simply faster computation. It is the possibility of designing molecules with a more detailed picture of how biological systems fit together. AlphaFold 3’s ability to model interactions among proteins, DNA, RNA, and other molecules pushes the platform closer to the kinds of questions medicinal chemistry actually has to answer.

That is why the move toward human trials deserves attention. It marks a transition from scientific infrastructure to clinical product development. Plenty of technologies improve how researchers understand biology. Far fewer survive the long path to becoming medicines.

For now, the announcement is best read as a pivotal step rather than a proven outcome. The company says it is close to the clinic, and the coming trials will reveal whether one of AI’s most celebrated scientific advances can begin delivering results in patients rather than predictions on screens.

This article is based on reporting by Wired. Read the original article.