Finding New Drugs Is Laborious. This AI Startup Is Skipping The Lab Work

Potential Sciences, a British startup that launched on November 5, is betting that physics-based simulation can replace much of the lab work in drug design. The company plans to build large-scale chemical datasets by using basic principles of physics to simulate interactions between atoms, then small molecules, then complex amino acids and proteins. Those datasets would then be used to train AI models and prediction systems meant to identify highly targeted drugs for specific diseases, bypassing costly and time-consuming lab experimentation.

Founder and CEO Hatem Helal framed the goal in terms of failure speed: "We want molecules that are never going to work in the clinic to fail fast," he told WIRED, adding that this requires better predictors of how molecules behave as they interact with human biology. The startup says it has seed funding in the tens of millions of dollars from SovAI, the UK government's AI venture capital fund, plus Twin Path Ventures and other investors, though the founders declined to disclose the precise figure or valuation.

The pitch rests on a critique of the current field. Helal says a large share of AI drug discovery labs build models on a single public database of 3D atomic structures, the Protein Data Bank, which researchers have grown over decades with experimental results — but which remains too small to train models advanced enough to reliably accelerate the search for viable candidates. He cast the problem as one of scale: of all the atoms in the universe, there are around 10 to the 60th power ways to assemble them into molecules potentially given to a human, so drug discovery is a search problem about how fast you can navigate and prune that space.

If the approach works, Helal says, it could speed the development of treatments across many diseases, and by reducing the cost of experimentation, make it economical for pharmaceutical companies to pursue drugs for rare conditions that have previously gone untreated. He also described an ambition toward one-off molecular combinations tailored to an individual's biology.

Skeptics note the hurdle is accuracy, not ambition. Peter Wolynes, a chemistry professor at Rice University, said he is sure the approach "will be an improvement," but added that protein engineering today is "at the level of the Romans, not of the Renaissance." The open questions are whether Potential Sciences can predict how well drug candidates bind to disease-related proteins as accurately as it hopes — a marked leap forward — and whether the tens of millions in seed capital funds the compute and validation needed to prove it. Industry figures like Demis Hassabis have long argued AI will eventually deliver momentous scientific breakthroughs, and startups like this are among the first testing whether drug discovery is that frontier.

keyFacts:[