The author reflects on how AI agents have fundamentally changed his work habits, leading to unprecedented productivity. He describes opening multiple agent tabs simultaneously while working, a stark contrast to his previous focused workflow. This constant multitasking, even for tasks unrelated to his current project, triggers a physical reaction of leg bouncing, indicative of anxiety or intense focus, now associated with agent use. He acknowledges the immense boost in productivity, with projects previously taking months now achievable in weeks, and ideas quickly turning into tangible first versions. However, this increased capacity leads him to initiate more tasks, filling the time supposedly saved with new projects.The author observes that this phenomenon mirrors research showing people voluntarily taking on more roles and working in fragmented intervals when AI tools are available. He compares his agent-driven workflow to a continuous feed, where unrelated tasks are initiated and completed in an order beyond his direct control. This constant task-switching results in "attention residue," where cognitive resources remain tied to previous tasks, hindering performance on new ones. The author realizes his entire day has adopted this feed-like structure, blurring the lines between work, rest, and distraction.He highlights a growing incapacity to perceive things holistically, contrasting it with a recent experience of deep, focused interface design. While acknowledging the creative stimulation of AI, he feels a loss of the ability to see the full picture, like an orchestra conductor managing various parts without playing them himself. The ease of experimentation with AI, while productive, is turning creation into a form of entertainment, making it difficult to ascertain when he is truly resting.The author concludes that AI enables a factory that never closes, where the limit on creation is his own willingness to initiate new projects, not time or budget. He notes that AI has facilitated the creation of entirely new, previously unimaginable tasks, such as building a personal analytics interface. Unlike previous technologies that infiltrated personal time, AI agents are integrated into the core work process, fundamentally altering the nature of deep, continuous work.