How We Review AI Scribes

AI scribes are the first AI tool most independent practitioners will adopt — and the category most likely to be sold to them by someone with a financial stake in the outcome. This page explains how we review them.

What we score

• HIPAA & BAA. Does the vendor sign a Business Associate Agreement? If not, the product cannot legally touch patient data in the US, and the review ends here.
• Integrative template fit. Most scribes ship SOAP templates designed for primary care. We score how well a tool adapts to functional-medicine intakes, chiropractic assessments, midwifery visits, and other non-conventional formats.
• Cost for a solo provider. Single-seat pricing, transparent or hidden. We publish what the practice actually pays.
• Data retention & training policy. Is patient audio retained? For how long? Is any of it used to train models? “No” is the only acceptable answer for the training question.
• Offline capability. Does the tool work in a basement office with bad Wi-Fi? Independent practices are not all on enterprise networks.
• Overall verdict. A short, plain-English judgment.

What disqualifies a vendor

A vendor that will not sign a BAA, that trains models on session data, or that refuses to disclose where patient audio is processed cannot be reviewed here. It goes on the Vendor Wall instead, with the reason published.

What we do not accept

No vendor pays for placement. No vendor sees a draft. No vendor’s PR team is consulted. If a vendor is owned by, partnered with, or otherwise tied to BTFOH’s publisher (Resonance Agency, FrequencyOS, or Content Studio), we disclose it on the page.

The first comparison goes live this week. Read it at /tools/scribe-review.