A project coordinator has a client email open and a half-finished reply. She wonders whether she can paste the thread into the AI tool to tidy her draft. It contains a customer name and a price. Is that allowed? She is not sure, so she opens the policy document, scrolls for two minutes, finds nothing specific, and messages a colleague, who says probably fine. She decides to write the reply by hand, and the hour that follows is a little slower than it needed to be. Nothing dramatic happened in that scene. No data leaked, no rule was broken, no tool failed. A single small question simply had no ready answer, so it cost a few minutes, a message and some low-level worry. That is what a missing default looks like. This matters because Stay Simple is about removing decisions we should not have to make. The same question, asked a dozen times a week across a team, is a decision tax that a single plain default could remove. ------------- Context ------------- Most workplaces have by now told people that AI is welcome. Culture Amp's 2026 AI at Work Benchmark, covering around 112,000 employees, found that 85 percent say they are encouraged to experiment, and 86 percent say they understand the risks. On paper, that sounds like a confident, well-informed workforce. But a separate 2026 survey of US workers found that 44 percent say their company has no clear AI policy, or they are not sure whether one exists. Encouraged, informed and unsure all at once. The gap between those two pictures is where the daily friction lives. When guidance is general, every specific situation has to be interpreted. Is this email too sensitive? Is this spreadsheet client data? Is a summary of a meeting recording acceptable? Each question is small, but each one needs a judgement call from someone who would rather be doing the work. We usually respond to this by writing a longer policy. That tends to make things harder, not simpler. A twelve-page document answers every question in principle and none at the moment someone needs it. People stop reading, then stop asking, then guess.