A recent article describes how productivity app maker Doist is deliberately reducing AI’s role in automation.
Rather than keeping AI in the execution loop, the company uses it to interpret intent and then hands execution to conventional software. The argument is simple: less AI can sometimes deliver more.
What struck me was not the novelty of this approach. I recognized a similar pattern I had already quietly adopted without fully realizing it.
When I use AI, I don’t spend all day prompting it to help me process work.
That may sound odd coming from someone who writes about artificial intelligence, participates in a Copilot change network, and regularly collaborates with AI systems. Yet the most valuable lesson I have learned is not how often to consult AI, but when.
The popular narrative suggests that maximum value comes from maximum use. My experience has been the opposite.
For me, AI is most useful at moments of complete uncertainty, contentious ambiguity, eager exploration and quiet reflection.
Engaging AI in these scenarios helps me pressure-test assumptions, challenge wording, identify patterns and examine ideas from different angles.
This deliberate leverage surfaces something seemingly invisible yet powerful. The value of working with AI is not uncovered where the work ends; it is where the thinking begins.
autom8 