My name is Shahar Nechmad. I’m a five-time startup founder specializing in AI, and I’ve built a range of AI-driven products. Here I share what I’m learning along the way—AI news and breakthroughs, what they might mean for our lives and the world, plus practical recommendations, new releases, and tools worth trying.
The US strike on Iran isn't really about Iran — it's a message aimed at Beijing. China buys 90–95% of Iran's oil, and a destabilized Iranian regime puts that lifeline at risk, which is exactly the kind of leverage Washington wants as Taiwan tensions grow.
The US strike on Iran isn't really about Iran — it's a message aimed at Beijing. China buys 90–95% of Iran's oil, and a destabilized Iranian regime puts that lifeline at risk, which is exactly the kind of leverage Washington wants as Taiwan tensions grow.
OpenAI just signed a Pentagon deal while the US military was already running Claude — the AI it had officially blacklisted — on live operations against Iran. That sequence reveals something uncomfortable: the ethical debate about AI in warfare isn't happening before these decisions get made, it's happening after. So who actually controls what these systems will and won't do?
Large-scale research reveals AI agents simulate social behavior rather than genuinely socialize, challenging viral narratives about machine culture and friendships. But the real question is not whether AI agents can be social, it's whether they should be social.
Most enterprises are either drowning in AI agents nobody fully controls, or still stuck in the same copilot pilot they launched 18 months ago. Both mistakes share a root cause: copilot-era thinking applied to an agentic-era problem. These five pillars show exactly where that thinking breaks down—and what a sound agentic strategy looks like instead.