AI Startup Skips Lab Work in Drug Discovery

Drug discovery still runs on a brutal math problem: screen thousands of molecules in a lab, watch most of them fail, and burn years and millions of dollars before one compound makes it to clinical trials. Potential Sciences, a British startup that launched on November 5, wants to cut that lab work out of the equation almost entirely. Instead of testing physical molecules against physical diseases, the company plans to simulate the whole process computationally, starting with basic physics governing how atoms interact, then scaling up to small molecules and eventually full proteins. The resulting datasets would train AI models to predict which drug candidates might actually work in the human body before anyone synthesizes a single compound.

The startup has already pulled in seed funding in the tens of millions of dollars, with backers including SovAI, the UK government's AI venture capital fund, and Twin Path Ventures. Hatem Helal, the company's founder and CEO, says the goal is speed: catching failures early, on a computer, rather than months into expensive lab trials. Whether that bet on simulation over experimentation actually holds up is the question hanging over the entire project.