AI Startup's Cancer Predictions Are Already Coming True

Jan de Vries · 2026-09-18

What if we could test billions of drug combinations without ever stepping into a lab? That's the wild premise behind Big Picture Bio, a London-based biotech that just emerged from stealth with a combined funding round of €2.55 million—roughly $2.7 million. And they're not just talking theory. Their AI has already called clinical trial outcomes with eerie accuracy. ### The Money Behind the Mission The funding breaks down into two parts. First, a €1.75 million pre-Seed round (about $1.85 million) co-led by Kadmos Capital and Exceptional Ventures, with participation from Gloucester Ventures and angel investor John White. Then there's €815k (around $860k) in non-dilutive support from Innovate UK's Investor Partnerships Programme. That's a mix of equity and grant money—a smart way to stretch early-stage capital. Remy Kesrouani, Managing Partner at Kadmos Capital, put it plainly: "Big Picture Bio is the culmination of that journey, applying a genuinely differentiated approach to AI to one of drug development's hardest problems: predicting how complex clinical trials will actually behave." ### The Founders Who've Been Here Before This isn't a first rodeo for CEO Dr. Kerstin Papenfuss and CTO Dr. Mark Hammond. They've worked together for seven years, most recently at Deep Science Ventures. Papenfuss built and led DSV's therapeutics team, founding 12 companies. Hammond co-founded DSV and grew its portfolio from €174k to over €1.16 billion. They know how to turn science into businesses. ### How It Actually Works Big Picture Bio has built a generative "world model" that simulates how tumors, immune cells, and surrounding tissue interact. The goal? Figure out which drug combinations—and in what sequence—are most likely to stop cancer from developing resistance. Here's the problem they're solving: a full pairwise screen of just 100 drugs at 100 doses would require roughly 50 million lab experiments. That's impossible. But their models can search computationally in seconds. Dr. Papenfuss didn't mince words: "Cancer is not one disease driven by one target, but we still develop drugs as if it were. Combinations are how we beat it – and with more than 900 billion of them possible, no lab on earth can test its way to the right ones." ### The Predictions That Turned Heads This isn't just a nice idea. The company says its predictive approach has already been tested against real clinical data. They correctly forecasted the failure of Regeneron's fianlimab trial. And ahead of the 2026 ASCO oncology conference, they posted 14 predictions—12 of which proved accurate. That's a hit rate that gets people's attention. > "This funding takes our first designed combinations out of the model and into the lab." — Dr. Kerstin Papenfuss ### What's Next The company will use the funding to take its AI-designed cancer combination therapies from computer models into the wet lab. They'll start with cancer and immune-mediated diseases but aim to move beyond trial-and-error combination testing altogether. Their advisory team includes senior experts from AstraZeneca, Exscientia, and PhoreMost—so they're not short on brainpower. If Big Picture Bio can deliver even half of what they're promising, it could change how we think about drug development. No more guessing. No more wasting years on combinations that were never going to work. Just smarter, faster paths to treatments that actually help patients.