Anthropic’s Dario Amodei and OpenAI’s Sam Altman once slipped out a single-page PDF laying out the 4 core AI missions they were chasing. The curveball: the last one isn’t a model—it’s an ecosystem, a graph where hundreds of multi-agent systems collide, cooperate, and evolve together. The four missions, in full: 1 → Unmask hidden AI: sweep news cycles, financial markets, and online games for fingerprints of someone quietly deploying a breakthrough system 2 → Forge coding agents: build an agent that can win programming contests by writing programs that write programs 3 → Use AI as the shield: protect computer systems from elite attackers by meeting AI-powered hacking with AI-powered defense 4 → Grow a living agent graph: link persistent agents that interact over time—learning, developing language, and jointly pursuing goals Amodei, Altman, and Ilya Sutskever drafted this during the ChatGPT build—and these missions still steer them now. That one-page document is essentially the blueprint behind today’s biggest AI labs, including Anthropic and OpenAI. Read it, save it, and then start building your first graph system ↓(CyrilXBT)