This post recounts how a small behavioral-health practice proposed an AI-assisted intake follow-up workflow, received a legal rejection, and rebuilt a narrower version centered on data minimization, named human review, and exception logging. The redesigned process produced modest time savings and a defensible audit trail.
This article details a narrow administrative AI pilot in a small behavioral-health practice, listing exactly what data and outputs were allowed, what was blocked, and why those boundaries made the workflow defensible.
This article describes a bounded workflow for using AI to draft internal communications in a small behavioral-health practice while ensuring that confidential and identifiable information never enters the model.
This article recounts an AI experiment stopped after one week, the early warning signs that triggered the decision, and the design lessons that improved later pilots in a small behavioral-health practice.