Noam Brown is one of the researchers who built o1. When he says he won't predict AI timelines beyond 3 months, that sentence deserves more than a scroll-past. A year ago, the researchers closest to the frontier felt comfortable projecting 12 months out. Now the same people won't commit to a quarter beyond the current one. Not because they've lost confidence in the technology. Because the pace has made honest forecasting nearly impossible. This is the part of the AI conversation that most business planning ignores. Roadmaps built on 2030 assumptions are being written by people who are more confident than the people actually building the thing. That's a problem. Not because long-term vision is useless — it isn't. But because the assumptions underneath that vision are expiring faster than the planning cycles built to review them. The practical shift: stop anchoring your AI strategy to a 3-year horizon and start running it like a quarterly operating decision. What can we build, test, and learn in the next 90 days? What do we need to be true for this to work? What breaks if the model we're building on gets replaced in six months? The teams that will compound the fastest aren't the ones with the most sophisticated 2030 roadmap. They're the ones that have built the muscle to reorient quickly when the ground shifts. It shifts every quarter now. Plan accordingly. Full YouTube video here: https://radi8.it/TXa7sFd #AIStrategy