Optimization has always been the name of the game. Whether we gather fruit, build cars, or provide a service, efficiency can make the difference between barely surviving and thriving. The World Bank about this dynamic.

So whenever I have some work to do, I wonder how I could automate that or at least speed that process up. Until recently, that usually meant searching for some new tool or learning about new concepts. But balance is crucial here: spending all one’s time optimizing means getting no actual work done.

AI (or AI agents specifically) has changed that, in ways we’re only beginning to realize. What previously took days or weeks may only take a few hours now. Mind you, agents are not perfect – but neither are we.

Given the insane speed at which this tech improves, it doesn’t take a crystal ball to see agents outperforming us far more often in the very near future. The independent AI research organization METR finds that the complexity of tasks frontier AI agents can complete on their own has doubled roughly every seven months since 2019.

That offers a huge opportunity: Trying to stay ahead of the curve (or at least up-to-date) means actually living in the future, with all its benefits and problems. Trying to implement agents everywhere could mean getting a significant efficiency advantage over those who don’t.

But delegating more work to agents also changes what doing the work itself feels like. Civilization IV designer Soren Johnson once wrote “Given the opportunity, players will optimize the fun out of a game.”. Original article. Now, work isn’t a game, but I’d argue that there is a related problem here: Optimization can create its own kind of unpleasantness. Waiting for an agent to complete a task is usually incredibly boring. I then often find myself trying to kill the time, doomscrolling on 𝕏 or watching videos on YouTube.

This, of course, creates another optimization problem.

How can we avoid waiting for our agents? How can we best automate everything? How can we stay sane doing so? I don’t have definitive answers yet. But seriously thinking about these questions now may itself give us an advantage.

So that’s me. Trying to optimize everything, hoping it will compound. Trying to keep a balance, making my own work ever more pleasant.