What is the 30% rule in AI?

The 30% rule is a popular shorthand for how humans and AI should split work: let AI handle roughly 70% of a task (the repetitive, pattern-based part) while humans keep the final 30% that needs judgement, empathy or accountability. Some people quote it the other way round; the number matters less than the principle that AI output should not go out the door unreviewed. In practice this looks like AI drafting the email and a person approving it, AI triaging the enquiries and a person handling the sensitive ones, AI preparing the report and a person signing it off. Businesses that automate 100% of a customer-facing task usually regret it; businesses that automate the 70% behind the scenes get most of the benefit with none of the reputational risk.