Everyone who starts padel makes the same mistake. The ball goes past you, toward the back wall, and you turn and chase it. You try to hit it before it touches the glass. It feels like the safe thing. It is almost always the wrong thing.
I did this for months. Tennis habits, I think. In tennis the wall is the end of the point. In padel the wall is part of the court.
Lately I see the same mistake somewhere else. In how I work with AI.
The glass gives you time
The first thing a coach told me was simple: let it come off the glass. Step back, turn your shoulders, wait. The ball hits the wall, slows down, and comes back to you higher and softer. Suddenly you have time. You can choose where it goes.
It sounds easy. It is not. Every part of your body wants to deal with the ball now. Waiting feels like losing control.
Working with an AI agent feels exactly like that. You give it a task, it starts writing code, and after ten seconds you want to jump in. Stop it. Correct it. Rewrite the prompt. Take the ball before it reaches the wall.
Most of the time, that is the wrong move. The agent is the glass. Let it finish. Let the ball come back to you. Then you see what you actually have, and you choose the next shot calmly.
Not every ball is an attack
The second lesson took longer. In padel you do not win most points by hitting hard. You win them when the other pair makes a mistake. The lob, the slow ball to the feet, the boring ball down the middle. Patience wins more than power.
With AI tools I used to go for winners every time. One giant prompt. "Build the whole feature." Smash. And then I spent an hour cleaning up the mess.
What works better is boring. Small tasks. Clear context. One ball down the middle, then another. Read the diff. Run the tests. Next ball. It does not look impressive, but the points add up.
Know your partner
Padel is two against two. Good pairs do not both run for the same ball. One takes the middle, the other covers the lob. You talk. "Mine." "Yours."
Building small products alone, I never had anyone to say "yours" to. Every ball was mine. Now I sort of do. The agent takes the tests, the boilerplate, the migration script, the first draft of the docs. I take the decisions: what to build, what to cut, what is good enough.
The mistake is to let it play your side too. Or to play its side for it. If I rewrite every line it produces, I am just chasing balls again, only now there are two of us on the same spot.
Practising the wait
What helped on court was boring drills. Someone feeds balls into the back glass, again and again, and you practise one thing: turn, wait, hit. After a while the body learns that the wall is a friend.
I am trying to build the same habit with AI. Give a clear task. Step back. Do something else for a few minutes. Come back and read what it did before I react. Not a rule, just a habit. Most of the time nothing breaks. Often the result is better than what I would have done in a hurry.
Padel is a strange teacher. It is a social game, a lot of laughing between points. But the most useful thing it taught me is very quiet, and it turns out it works off court too. You do not have to take every ball before it reaches the wall.
Anyway. See you on court. Or in the terminal.