We hate to wait. While an LLM generates an answer, we impatiently open another tab or terminal and start a new task. Then another, and another.

This is an unhealthy pattern of human-AI interaction. It drains our mental juice faster than the back-to-back unwanted meetings we were all complaining about in the pre-AI era :)

On the other hand, I’m happy to see healthier workflows emerge. One of them is The Day Shift / Night Shift.

During the Day Shift, humans are doing enjoyable activities: planning, big-picture thinking, designing feedback loops, and other creative and human-enjoyable activities.

During the Night Shift, the AI executes the plan produced during the day. The next morning, we review and QA the results.

While this workflow won’t apply to every single task, and many things can go wrong, I’m happy to see that it’s emerged at all. Many more will emerge.

Imagine a workflow in which the AI focuses on maximizing your creativity. It knows which activities can spark your inspiration, and when you hit a wall, it simply suggests: “Go take a walk in your favorite park. Grab a notebook and capture creative ideas. I’ll be here when you are back.”