AI Pulse · Frontier Labs Watch

Google just disbanded the team that built AlphaFold, the protein-folding model that won a Nobel Prize last year. The researchers are being reassigned to Gemini and broader science initiatives. Engadget reported the move signals a shift from building the best narrow model for a specific domain to one general system that handles everything.

This matters because it shows where the big labs think the value is. A few years ago, the story was specialist models. AlphaFold cracked a decades-old biology problem. It was celebrated, prestigious, and scientifically important. But Google is now betting that a single general-purpose model, trained on everything, will be more useful and more defensible than a stable of domain experts. That changes how you think about adopting AI in your business. If the frontier labs are consolidating around general models, you probably don’t need to commission a bespoke model for every workflow. You need to get good at prompting, structuring data, and connecting a capable general system to your actual work. That’s the kind of thing we build into an AI command centre at Omni, where one interface routes tasks to the right model and keeps your team moving without needing a PhD.

The talent angle is worth noting too. AlphaFold’s lead researcher, John Jumper, left for Anthropic in June and brought colleagues with him. That’s a direct pipeline from Google’s Nobel-winning work to a competitor. It tells you the best people are mobile, and they’re choosing labs based on research direction and autonomy, not just prestige. For business owners, the takeaway is simpler. The era of waiting for a perfect specialist model is over. The general models are here, they’re improving fast, and the question is whether you’re set up to use them.