Earlier this week at Machines Can Think Summit 2026, our keynote was watched by 1,000 people in the room and over 4 million live on Twitter. The topic was ambitious but concrete: how AI can finally make drug development faster, cheaper, and more predictable.
In the talk, we shared Bioptic’s journey — from training AI models, to synthesizing molecules, to building AI agents that run the business of pharma. The core insight is simple: while breakthroughs like AlphaFold matter, science is only the tip of the iceberg. Most drugs fail not because the molecule is wrong, but because decisions around competition, regulation, clinical strategy, and commercialization are made too late or with incomplete information.
That’s why Bioptic evolved from an AI-for-science company into an AI-native operating system for pharma decision-making. Our agents don’t replace scientists or executives. They give teams anticipatory intelligence across science, business, and regulation — so better decisions get made earlier, when they matter most.
The full keynote video is embedded above. If you’re curious how we think about vertical AI, trust, evals, and why “boring, practical agents” may matter more than sci-fi AGI, it’s worth a watch.
.png)