Deterministic, not improvised: why an LLM should never decide care

By Tom Higgins — I run Zvia, my Lakewood medspa, on the tool behind this.

There’s an important difference between an AI that drafts and an AI that decides. Blurring the two is how clinical software gets dangerous.

Code decides; AI drafts

In ProtokolIQ, treatment protocols and contraindication screening run on a deterministic engine — explicit rules a clinician can read, test and audit. There is no black-box language model quietly deciding whether a treatment is safe. AI drafts the note, the campaign, the summary; a licensed provider decides and signs.

Why determinism matters for care

Auditability. A deterministic rule produces the same result every time and can be explained line by line.
Accountability. The human who signs remains accountable — the system supports judgment rather than replacing it.
Safety. Prescribing and dispensing decisions are human-only by design.

The trust ratchet

Where AI does assist, autonomy is earned, not assumed. Every AI action sits on a five-level ratchet — from “show your work” to “act and surface only exceptions” — with promotions earned by demonstrated accuracy and demotions automatic on error. Read how it works.

When you want to see where your own practice stands, the Clinic Freedom Score walks you through it in a few minutes — or see the platform. Not a minute before.