Fragmentation beneath the flow is the distinctive cognitive pathology the simulation attributes to AI-mediated work. The engineer who spends three hours in continuous conversation with an AI tool experiences sustained engagement — a flow state by every subjective measure Csikszentmihalyi would check. But the underlying cognitive reality involves rapid transitions across distinct domains: product thinking, systems architecture, implementation review, quality assurance. Each transition carries attention residue; the conversational continuity conceals the transitions; the worker accumulates cognitive cost without the metacognitive signal that would normally register the cost as depletion. The concealment is the distinctive contribution of the AI interface — the first technology in the escalation sequence that fragments attention through integration rather than through interruption.
The mechanism differs categorically from the email-era fragmentation Perlow originally diagnosed. Email interruptions were at least partially salient. The worker who was pulled away from a task could feel the break; the friction of the interruption served as a signal. AI interaction eliminates the signal. The transition from product thinking to architecture review does not feel like an interruption — it feels like a deepening, a natural progression of the conversation. The worker does not experience the cognitive cost of the domain switch because the tool's seamless interface presents each switch as continuity.
This creates a diagnostic problem that Perlow's original framework does not address directly. Her BCG interventions assumed that workers could identify the interruptions fragmenting their attention, because the interruptions had discrete sources — email, phone, Slack message — that could be counted and potentially regulated. AI-era fragmentation resists this identification. The transitions are not interruptions from the tool's perspective; they are features of its design. Regulating them would mean regulating the tool's core functionality. The fragmentation is woven into the productivity rather than imposed upon it.
The implication is that the unit of intervention must shift from managing individual interruptions to limiting total session duration. The team cannot agree to reduce the number of domain switches per conversation, because the switches are not experienced as switches. The team can agree to limit the duration of continuous AI engagement, creating structured boundaries that interrupt the accumulation of residue before it reaches the threshold of cognitive impairment. This is the operational pivot that AI-era applications of Perlow's framework require — from managing the inputs that fragment attention to designing the containers within which engagement happens.
The concept is developed explicitly in this volume's simulation, drawing on Perlow's framework combined with Sophie Leroy's attention-residue research and the Berkeley study's empirical findings on AI-era workflow fragmentation.
Integration, not interruption. Fragmentation is woven into the productive workflow rather than imposed from outside it.
Invisible transitions. Domain switches feel like conversational depth, not like task-switching — defeating the worker's capacity for self-monitoring.
Cumulative residue. Individually imperceptible residue layers accumulate across hours into measurable cognitive depletion.
Shift the unit of intervention. Regulate session duration rather than interruption frequency, because the transitions are not countable in the old sense.