Three loads, one outcome
Three kinds of load pile up across a working week. When recovery doesn't keep pace, capacity leaks: the slow drain that never shows up in output metrics until performance has already moved.
Cognitive load
The mental demand of processing information, deciding, switching tasks, and holding complexity. When load exceeds capacity, clarity drops, errors rise, and creativity is the first thing to go. Switching leaves a trace, too: part of your attention stays on the last task after you move to the next. The name for that trace is attention residue.
Emotional residue
The emotional carryover from meetings, interactions, and hard calls. Unlike ordinary tiredness, it doesn't resolve with rest alone. It needs to be actively processed, which almost no workday makes room for.
Contextual friction
The resistance you feel when tools, workflows, and norms get in the way of applying your capacity: digital friction, coordination drag, and the newest form, the burden of supervising AI output.
Patterns we study
The doom loop
Overload leads to reduced capacity, which means more hours are needed, which means less recovery, which deepens the overload. The cycle compounds quietly until it looks like a performance problem. It isn't one. It's a design problem.
Capacity erosion
The slow wearing-down of usable capacity over weeks. It reads as declining performance or fading motivation, but it's a state change in the person, produced by the design of the work around them.
The field we're building in
We work at the intersection of cognitive load theory, occupational health psychology, and human factors, and we're honest about what's established science versus what's our own working practice. Two questions we think are genuinely new: how the burden of supervising AI hides inside fluency ("it's fine, I just double-check everything"), and what happens to attention in the 10-to-120-second gaps while people wait on AI output. If you're a researcher interested in either, we want to talk.