Analysis
What the numbers actually mean.
More than half of everything typed into Active Intelligence is a unique, complex request, while every template-friendly category combined barely moves the chart. That tracks with the largest study of consumer AI use to date, where OpenAI found nearly half of all ChatGPT messages are people asking for answers and judgment rather than running a canned task.
Report requests, short prompts, and recommendations each sit under 5%, so a tool built around pre-set buttons would be optimizing for behavior almost nobody exhibits. The value lives in the long tail, in the specific, one-off questions no template could have anticipated.
If the highest-value behavior is custom questioning, then what separates strong AI users is the quality of the question they bring. Marketers who get specific about their actual problem pull far more from the same tool than anyone clicking through canned suggestions.