About Us

About Us – The Defensible Practice

I am Daphne Grimaldi. I live in Columbus, Ohio, with my partner and our school-age child. For years I worked inside a five-to-fifteen-person behavioral-health practice, building its compliance and operating system from the ground up—policies, vendor reviews, staff training, patient-communication routines, incident procedures, and the day-to-day discipline that keeps a small clinic from becoming a liability the moment something goes sideways. When the practice wanted to introduce AI, I lived through the entire uncomfortable sequence: the business case looked promising, legal and privacy concerns stopped the first attempt, the proposed workflow was rejected, and we had to redesign it around data minimization, human review, access controls, documentation, and the real capacity of people who already had full caseloads.

That experience is why this site exists. Not because I collected a set of best practices from conferences or vendor webinars, but because I had to sit with the gap between a tool’s marketing page and the Tuesday-afternoon reality of a small practice that cannot afford a hospital-sized legal department.

Building the System Before Anyone Asked for AI

Most of the work that prepared me for AI decisions had nothing to do with artificial intelligence. It was the quiet accumulation of operational discipline: writing procedures that a busy clinician could actually follow, designing vendor questionnaires that forced real answers instead of marketing language, keeping training records current when staff turned over, and learning the difference between a policy that looked responsible on paper and a process that survived contact with real patients and real schedules.

I still keep a paper notebook of open questions, decisions, and follow-ups. Memory is unreliable when the stakes involve patient trust and regulatory exposure. The notebook is not romantic; it is practical. It is also the same habit I later used to document AI pilots—what we allowed, what we blocked, where the first workflow failed, and what human review actually required once the enthusiasm cooled.

The First AI Attempt and the Redesign That Followed

When the practice first proposed an AI-assisted workflow, the conversation moved quickly from “this could save time” to “this creates risk we cannot yet explain.” Legal and privacy concerns halted the initial version. We stepped back, narrowed the scope, and rebuilt around explicit data boundaries, separate accounts, documented human review points, and a one-page decision record that named who owned the judgment. The tool never signed the note. A person did.

That redesign taught me something I now repeat often: slow is not the same as behind. A six-week pilot that produces a defensible process is more useful than a two-week launch that leaves everyone guessing what to say if a patient, an auditor, or a licensing board asks what happened.

Running, Cooking, and the Shape of Careful Work

Outside the practice I train for long-distance races. Long runs are where I think through operational problems that do not resolve in a meeting. The same patience required to hold a steady pace over twenty miles shows up when a practice is deciding whether an AI use case is ready or still too vague to approve. I also cook most weekends—recipes that reward preparation without pretending the kitchen is a laboratory. Both habits reinforce the same editorial rule I try to follow here: do not promise shortcuts around judgment, verification, or professional responsibility.

I live in Columbus with my partner and one school-age child. The family schedule is ordinary and full. That ordinariness matters. The people who read this site are usually balancing clinical work, staff management, and the rest of life. Guidance that ignores that reality tends to be useless by Wednesday.

What This Site Is For

I write for owners, practice managers, operations leads, privacy staff, and clinical leaders at small behavioral-health organizations—and for the secondary readers in dental, physical therapy, home health, and other regulated specialty practices who face similar constraints. The site is not a product catalog. It does not sell certainty or claim that any article, template, or vendor statement makes a practice compliant by itself. Clinical decisions and patient-facing documentation remain under appropriate professional and organizational review.

The five categories—The Judgment Call, Workflow Under Review, The Small Practice Stack, Notes You Can Defend, and The Long Run—are the shape of the work I keep returning to: deciding whether a use case belongs at all, testing narrow workflows with real boundaries, building the operational stack before any tool is tested, recording decisions so they can be explained later, and paying attention to the human pace that makes any new process stick.

If you are looking for a list of the ten best AI tools or a promise that cautious practices are falling behind, this is not the place. If you need a careful answer that treats the details as consequential, you are closer to the right neighborhood.

— Daphne Grimaldi

The Defensible Practice

Columbus, Ohio

Last Updated: 2026-11-16 13:18