On a Thursday afternoon, an operations coordinator finishes a tense call with a supplier. For years, the hour after a call like that was gentle. She filed the invoices, tidied the shared tracker and answered the three simple emails. Nobody called it a break, but her shoulders came down a little with every one. This week, AI did all three before she hung up. The next item in her queue is a budget dispute with a colleague she likes. She starts it with the supplier call still ringing in her head, and no gap in between. That scene is becoming ordinary. In BCG's 2026 survey of nearly 12,000 workers, 67% say AI has taken over their simpler tasks and left them with more complex work, and 41% report more mental strain. This is a Stay Human issue as much as a workload one, because the people we are at our best are rarely the people who have been working hard for four hours without a pause. ------------- Context ------------- When we talk about what AI takes off our plates, we usually count the hours. Filing, formatting, first drafts, meeting notes: gone, and good riddance. Nobody misses them as tasks. But easy work was never only about the output. It also set the rhythm of a day. A hard conversation was followed by something light, a stretch of deep thinking by a few mechanical jobs, and our minds got the chance to settle, drift and reset before the next demanding thing arrived. That rhythm was never written into a job description, so nobody noticed when it started to go. The simple tasks were doing a second job in the background, and it was a human one: recovery. We did not schedule it. It simply came with the work. Now the easy layer is thinning out, and what remains is the hard layer, one item after another. The risk is not that the work is too difficult for us. It is that the qualities we bring to hard work, patience, judgement and warmth, run on energy that the easy tasks used to help us refill. If we want AI to leave us more human at work and not less, we need to notice what was quietly holding that up.