# Frontier Dental > Expert dental knowledge, built where models fail. We make dental AI reliable, safe, and effective in clinical settings. Our expert network writes evals to improve AI use with gold-standard answers. Frontier Dental is a Toronto company. Contact: hello@frontierdental.io ## Proof - Written by practicing dentists - 0 of 12 frontier model answers passed our sample case - No patient data: every case is a composite Full sample case: https://frontierdental.io/case-study.md ## What's in a case A case is a clinical question, a gold-standard answer, and a rubric that scores any LLM answer against it. - **The question.** Written by a practicing dentist in clinical language. - **The reference answer.** A gold standard, written by the dentist in complete consideration of all patient factors and treatment options. - **The rubric.** A measurable score for LLM answers across criteria. Each criterion checks one thing, carries points or a penalty, and comes from the dentist. - **Graded model answers.** Every answer from each model, scored line by line against the rubric. Cases are tagged by dental specialty and by the type of clinical work. Today's cases are treatment planning questions, in text. Use them to measure your models and improve them in post-training. ## How a case is made Dentists write the cases. We keep only the ones models get wrong. 1. **Write.** A practicing dentist writes a clinical question where the outcome depends on patient context and expert knowledge. Often this involves a health factor, a subtle choice between comparable options, or a pitfall that may be unsafe. The same dentist writes a gold-standard answer and a rubric that scores any answer against it. 2. **Test.** We give the case to leading models, with set criteria for model difficulty. This could be a pass rate, a minimum score, or another measure of success our customers choose. Each model solves the case several times, to show that its knowledge gap is real. These solutions are the model attempts. 3. **Grade.** Our rubric defines the standard set by our expert network. Each item lists what a good answer should say, with penalties for unsafe or evasive replies. An LLM-as-a-judge acts as the grader and applies that standard across all the model attempts. No model grades its own answer. The result is a score for how well each model answered. 4. **Keep.** When a score is below a set threshold, we deem the case difficult. Our rubric also exposes specific gaps, such as treatment blind spots or safety pitfalls, which point to a valuable direction for training. Those gaps are what our customers use our product for: to measure and improve their models. ## Pilots Two ways to start. ### Case Pilot 25 cases. Delivered in 1 month. - Model-hard on the models you name, against the pass line you set - Spread across the specialties you choose - Question, gold-standard answer, rubric and graded model answers for every case - Tagged by specialty and type of clinical work ### Eval Pilot 50+ cases. Your dental AI, tested on dentist-written cases. - You choose the specialty and clinical workflow - Your product answers a chosen set of cases - Every answer graded against the dentist's rubric - A report showing where it misses the plan a dentist would choose ## Questions ### Who owns the cases? Frontier Dental owns the cases it produces. Licence terms, including who else may receive a case, are agreed per project. ### Is there any patient data? No. Every case is a composite written from a dentist's experience. No names, dates or details that could identify a real patient. ### Who grades the answers? Dentists set the standard. Model graders apply it, line by line. On the sample case, each answer was graded by the two other models. A line counts in full when both graders agree it is met, and half when only one does. ### Can we choose our own models and pass line? Yes. You set the bar for your project. ### How does a pilot work? 25 cases, delivered in 1 month, spread across specialties. ## Request a sample pack Five dentist-written eval cases, right in your inbox. Enter a work email in the form at https://frontierdental.io/#sample-pack, or email hello@frontierdental.io.