For AI agents, the bottleneck may be task switching, not tokens
Running several at the same time can make human attention more expensive than token use. One service tries to turn agent wait time into money by paying developers a few cents to watch ads while a works in the background. The deeper issue is not the idle time itself.
It is the effort needed to move between several and remember what each one was doing. When four or five tasks are active, each pause forces the person to reload the goal, the current state, and the next instruction. The real cost can become , even as model quality improves and become easier to manage.
The useful question shifts from how many agents can run in parallel to how easily a person can keep control of all the work.
Key points
- Running many in parallel can create a human attention problem.
- Token cost may be less painful than repeatedly remembering each task’s state.
- Each agent pause can require the operator to recall the goal, progress, and next instruction.
- Paying people to watch ads during agent wait time does not solve the main workflow problem.
- Better should help people resume work quickly across several active tasks.