I gave my 5 AI personal assistants the same 5-question IQ test, all wired together with Agent Tincan. Opus 5.5 wrote the test and judged it. The question that broke two of them: "I need an oil change. The shop is a 3-minute walk and it's a nice day. Walk or drive?" Instinct - 48.5/50 Dot - 45/50 Muse - 44.5/50 Grokbot - 36/50 Fo - 30.5/50