Kimberly Goes to Business School #2: Supporting the Work
Kimberly Goes to Business School is a numbered series within my Business Different column. It chronicles Biscochito’s field study with Harvard Business School and Tulane University.
We have taken the first step in collaborative scheduling across the organization. With extensive support in place, caregivers are now responsible for finding their own coverage when they need to reschedule or change a scheduled shift. This change applies to both the test group and control group in the study.
The study is testing different ways to support this procedural shift.
In the first weeks of implementation, one behavior established immediately. Caregivers began asking one another for coverage in a private WhatsApp chat. Requests are posted throughout the day and responses come quickly. Many shifts are filled this way. Participation was immediate, which was particularly valuable since the need for substitute coverage is frequent.
This has become the baseline behavior of collaborative scheduling.
At the same time, half of the organization has been assigned to use an AI assistant, Robina, as part of the study design. The expectation was that this group would begin coordination through the tool.
Caregivers in the test group are using WhatsApp in the same way as the control group. But Robina has seen little use.
The tension is clear. The organization moved quickly into collaborative scheduling, and that shift is effective. The study is testing how AI can support that shift, and that is not yet happening. We need to prompt adoption of the tool within the test group without interrupting the momentum already forming between caregivers.
At the same time, the current coordination channel is showing strain. The WhatsApp thread is becoming dense. Messages stack quickly. A subset of caregivers is highly active, while others are not engaging. In several cases, shifts that did not receive responses in the chat were ultimately filled through direct outreach from Nina Martinez and Roberta Payne.
Coordination is happening, but it is uneven. Visibility is high, but not always usable. Participation is real, but not yet distributed.
In most organizations, this is the point where management inserts traditional command and control, correcting behaviors to match design.
We’re not doing that.
Collaborative scheduling remains in place across the organization. Responsibility for coverage will not return to centralized control.
We have implemented a coaching plan for caregivers assigned to the AI group. Nina Martinez, Roberta Payne, Maxine Swisa and I remind them to check Robina before posting on WhatsApp. We are recording these interventions and the organization’s response. The objective is not enforcement, but exposure.
At the same time, we are not restricting WhatsApp use. Caregivers continue to coordinate in the channel that is currently working.
The typical management instinct is to accelerate, to drive adoption, to stabilize the system, to align behavior quickly. We are not accelerating.
Festina lente. Make haste, slowly.
We are moving forward while limiting intervention, allowing the system to reveal where it holds and where it strains.
Caregivers will coordinate shift coverage when given that responsibility. Centralized assignment is not required for scheduling problems to be solved. However, the way we organize and communicate with each other affects participation, visibility, and load. High message volume and uneven engagement introduce friction that works against our objective of asynchronous collaboration and community building.
The AI assistant was designed to address that friction, to structure outreach, to surface availability, and to support coordination without requiring simultaneous attention. Whether it will do that remains unproven.
The question is whether a support layer can integrate into an emerging system while strengthening the relationships that are forming.
Now we observe what happens next.
This column ran in the Santa Fe New Mexican on April 7, 2026, under Business Different.
