The case,
pre-screened
before you open it.
Three steps from upload to verdict.
- 1Intake
Upload
STL scan, panoramic x-ray, intra- and extra-oral photos, and a short intake form. Drag and drop in the browser.
- 2Engine
Romeo pre-screens
Mild, Moderate, or Difficult, plus any override flags the engine raised against the rubric: impactions, skeletal discrepancy, perio gates.
- 3Report
Open the report
Verdict card, twelve-category breakdown, cited rationale, and a Case Companion chat for follow-up questions on the case in front of you.
A second opinion, grounded in the math.
The verdict comes from a deterministic, versioned scoring engine. The model reads the case and explains the result; it doesn’t classify it. That separation is the whole point.
Every line of rationale points to a subscore, a rubric clause, or a look-alike case in the Evenly library. Inputs, prompts, model versions, and engine outputs are written to a permanent audit record on every case.
Romeo opines like a senior orthodontist would — direction of treatment, what to watch, which additional records to pull — with the math and the evidence visible underneath.
- Categories
- 12 rubric subscores
- Calibration
- 20,000 Evenly cases
- Audit
- Append-only, per case
- Surface
- Server-side only
“The verdict is math. The model’s job is to read the case and explain itself, not to decide.