Did it work?
Short case studies for practice owners weighing AI tools. Each one separates claims, what we checked, and limits, and links every source. Sorted by strength of evidence.
Atmosphere · luis gomes / Pexels · not a clinic photo.
- Level 4Randomized trial
Radiograph AI for caries: more lesions found, and more drilling?
A randomized trial found AI support raised dentists' sensitivity for enamel caries on bitewings, and also raised invasive treatment decisions. We set that next to two FDA clearances and the vendors' own launch numbers.
Partly. Better at finding early enamel lesions in one trial, with a real risk of more invasive treatment.
imaging · caries · FDA 510(k) · updated 2026-10-31
- Level 3Organization evaluation
Ambient AI scribes: one year and 2.5 million uses, in medicine rather than dentistry
The Permanente Medical Group reported a year of generative-AI scribe use across 7,260 physicians. It is the largest public rollout we found, and it is not a dental one.
For these physicians, mostly yes: less after-hours charting, and heavy users gained the most. The evidence we read did not test transfer to the dental chair.
notes · charting · scribes · updated 2026-10-31
- Level 2Operator-reported
DSO AI ROI: phones show-rate lift and revenue-cycle productivity
Operator-reported results from a Becker’s Dental Review roundtable: AI phones (~7% show-rate lift) and revenue-cycle AI (roughly doubled coordinator productivity).
Directionally encouraging. These are two operators' own numbers from a trade roundtable, with no baseline or denominator published.
phones · revenue-cycle · DSO · updated 2026-10-31
- Level 1Vendor case study
AI phones after hours: what the vendor case studies actually measure
Two dental AI-phone vendors publish customer stories with booked-production and revenue figures. We read both line by line to separate measured counts from projections.
Unclear. The call and booking counts are believable signals. The dollar figures are vendor-calculated, with no baseline, no control period, and no cancellation data.
phones · front desk · after-hours · updated 2026-10-31
The evidence ladder
A higher rung is stronger evidence about a claim. It is not a product rating, and it is never sold.
- 1 Vendor case study Written and published by the company selling the tool. Useful for what to ask; not proof.
- 2 Operator-reported A practice or group leader describing their own results, usually without a method section.
- 3 Organization evaluation A health system or group measuring its own rollout at scale, often published, rarely randomized.
- 4 Randomized trial Clinicians randomly assigned to use the tool or not, with an independent reference standard.
Full definitions: editorial policy. Spot an error? Corrections.