The thirty seconds before the prompt

I had a very good week recently. Documents that used to take an afternoon were done before lunch. Analyses came back structured and reasonable. Nothing had stopped working. Everything had got easier.

That is the part worth sitting with. Fatigue announces itself, and you can usually name the meeting that emptied you out. What I notice in myself is quieter than that. When I stop forming a view before I prompt, the habit gets thinner, and the change does not arrive as loss. It arrives as relief.

What should catch this is metacognition: how we monitor our own understanding, and how we steer what we do next. Difficulty is usually what tells me to slow down and check. When the answer arrives fluent, that cue does not arrive with it.

Steven Shaw and Gideon Nave at Wharton have a name for the move underneath. They define cognitive surrender as adopting AI-generated outputs with minimal scrutiny, letting artificial cognition override both intuition and deliberate reasoning.

The word I keep coming back to is surrender, and that part is my own reading. Surrender is not defeat. It is something you do voluntarily, usually for decent reasons, often with a sense of relief. Which is why I would answer it with a practice.

McKinsey published a good list this month, five principles for what they call brain-powered organizations. Calibrate cognitive load. Protect cognitive capacity. Enable focus. Build adaptive brain skills. Drive brain-positive environments. Every one of them does real work, and two of them keep working before anyone has spotted a problem.

Adaptive skills are the first. Training helps. Skills of this kind are also kept by use, so the exercise is small and daily: do the thinking while the shortcut sits there, available, and take the shortcut second.

Environment is the second, and it is the one leaders can build on purpose. If this is hard to feel from the inside, the people best placed to notice are the ones working next to you. That means a team where “I don’t think you had a view on that” lands as a favour. It is worth designing, and it is where I would start.

My own practice, imperfectly held. Before opening the chat, decide what I think. Thirty seconds. One line if it helps.

Not because my answer will be better. Often it is not. But with a view of my own I am reviewing a proposal rather than reading a conclusion, and the model has something to push against.

The team version is the same move with a rule around it. On decisions that carry consequences, everyone brings a position before the system sees the question, and the review afterwards asks what changed and why.

That is the version worth aiming at. Keep the speed, and keep the first move in judgment human.

Curious what works for other people here. What habit helps you form a view before you prompt?