Detection is not the same thing as benefit

Dental AI is no longer experimental. In 2026, 43.3% of US dentists reported using it for at least one task. But fewer than 5% used it for treatment recommendations, and three in ten said they never intend to use it at all.

That caution is well founded, and the reason sits in the evidence rather than in temperament. Higher sensitivity applied to a screening population produces more findings — and whether those become better care or simply more restorations depends entirely on the treatment threshold applied afterwards.

The health-economic modelling that found AI cost-effective was explicit that this only holds where detected early lesions are managed non-restoratively.

Which points at something most vendor conversations skip. AI does not change your treatment philosophy. It multiplies whichever one your practice already has.

The Intelligent Practice

Six parts, 27 chapters, in English, Spanish and Dutch — identical chapter and protocol numbering across all three, so a multi-site or multilingual team works from the same checklists.

Clinical evidence Caries detection, bone-level measurement, CBCT segmentation and cephalometrics: what was measured, in whom, with what design.
Design and the lab Crown design, smile simulation, implant and aligner planning, and where generative systems reliably fail.
The practice engine Where the returns actually are in 2026, and why they are administrative before they are clinical.
Implementation A 90-day protocol, a vendor scorecard, 24 questions to send in writing, and a return model you can defend.
Risk and regulation FDA clearance, MDR Rule 11, and the EU AI Act after the Digital Omnibus, including the deadlines that reach your practice.
Tools Checklists, patient scripts, a twelve-metric dashboard and a twelve-month roadmap.

The standard every claim is held to

29 references, each carrying a DOI, PMID or PMC identifier.
Twelve of them, including every load-bearing figure, re-verified against source in September 2026. The verification record ships with the book.
Where the evidence is weak, thin or contradictory, the text says so.
Where a figure comes from market research rather than peer-reviewed work, it is labelled as such.
No commercial relationship exists between this publication and any vendor named.

Get the review

All three language editions are included in every tier.

Single reader €89 One clinician. Launch price for the first 50 copies, then €129.

Practice licence €390 One location, up to ten people, with the right to print the checklists internally.

Group licence €1,490 Multi-site, unlimited clinicians, including editable checklist and dashboard files.

Buyers receive free updates within edition 1.x and half off edition 2.0. Read it, and if it does not change how you would approach your next vendor conversation, email within 30 days for a full refund.

Get the review Buying for a group? Email me

Detection is not the same thing as benefit

The Intelligent Practice

The standard every claim is held to

Get the review

Not ready? Start with the vendor scorecard