AI that showsits working.You stay the prescriber.
Two assistants, and neither runs until you ask. Case Analysis turns a case in plain words into real Kent rubrics you edit, then ranks remedies by arithmetic you can open. AI Chat answers from the classical library first — and tells you when it is not.
Real Kent rubrics · Open arithmetic · No potency, no diagnosis
The leader covers most rubrics and is well clear of the next remedy.
Decision support only. The prescription is the practitioner’s.
The AI picks rubrics. It never picks the remedy.
AI is used where it helps — reading a narrative and choosing from real rubrics. The ranking is arithmetic, and every number in it can be opened.
Plain words in, typed symptoms out
Type the case the way you heard it. Analyze splits the narrative into mental, causation, general and particular symptoms, each with an intensity. Five example cases are there to try first.
Fenced to real Kent rubrics
Candidate rubrics come from Kent’s repertory, and the AI may only choose from that menu. Anything it invents is dropped — and the number dropped is shown to you, not hidden.
You edit the rubrics
Remove a rubric, add one by repertory search, or mark it as more important — the strange, rare and peculiar of §153, counted 1.6×. Nothing is ranked until you press Repertorize.
Arithmetic you can open
The ranking is not AI. Grade × symptom class × intensity × a small-rubric factor that stops polychrests dominating. “How this was calculated” shows the working for any remedy.
Confidence, stated honestly
A High, Medium or Low band, explained in words — agreement within the rubrics, not a probability of cure. When the top two are near-tied, a differential pair banner says so.
Guardrails, not promises
Fewer than three rubrics is “Case too thin” — no ranking. No remedy outside your rubrics, no potency, no diagnosis, and a disclaimer that stays on screen and on print.
A colleague who cites the books
A homeopathy-only assistant. It reads the library before it answers, fills gaps from general knowledge only when it must, labels that part, and lets the library win on any conflict.
- Library first — Boericke, real repertory rubrics, Organon aphorisms when a § is cited — with general knowledge labelled as such
- Two reply modes: Case-taking asks structured questions; Quick shortlist gets to the point
- Find a rubric and get real rubric rows, with add-to-case and a Materia Medica overlay
- Sees the repertorization case you have open
- History synced across devices — auto-titled, pin, delete
- Replies in your interface language (19), remedy and rubric terms kept in English
- Voice dictation, best in English
- Print a conversation as a consultation, with the disclaimer
Two real Kent rubrics fit what you have described:
Does the weeping ease with consolation, or is it worse for it?
General knowledge — labelledCase → Rubrics → Result, then the record
Three steps you can see, each one yours to change — and a case that lands where you will look for it next visit.
Case
Type the narrative and press Analyze. Symptoms are typed and weighted; Kent rubrics are proposed. If AI is unavailable, rubrics are still proposed without it — and labelled so.
Rubrics
Remove, add by search, mark what is strange, rare and peculiar. The list you leave is the list that is ranked.
Result
Up to 80 remedies (20 by default) as a grid or list — Match %, coverage, the working, a confidence band, filters, and a link into Materia Medica for each.
Save
Save the case to Saved Cases — and to the patient’s Encounters when a patient is active. The session is kept per patient; print carries the disclaimer.
On the result, Explain this result writes a reading of the ranking on request, and Ask about this result can suggest a rubric — matched to a real Kent rubric and added only when you tap Add rubric, after which the ranking is recomputed.
AI only when asked
Nothing calls the AI when a page loads, a tab opens or you print — only when you press Analyze, ask for an explanation or send a chat message. Each clinic has a daily AI allowance that resets at local midnight.
Validation note: on 15 textbook cases, the textbook remedy appeared in the top three every time and ranked first in 11 of 15. Textbook cases are cleaner than a consulting room — treat this as a check on the method, not a promise about your patients.
Questions, answered
Does the AI choose the remedy?
No. In Case Analysis the AI only helps pick rubrics — and only from real Kent rubrics offered to it; anything else is discarded and counted. The ranking itself is plain arithmetic you can open and check, and the chat is refused any remedy that is not on your ranking.
What do I have to type?
Just the case in plain words. CuraSim splits it into mental, causation, general and particular symptoms and proposes rubrics. You remove, add or mark any of them as strange, rare and peculiar before anything is ranked.
What does the confidence band mean?
How strongly the ranking is supported by the rubrics in front of it — nothing more. It is evidential agreement within the sources, not a probability of cure. When the top two remedies are nearly tied, CuraSim says so instead of picking one.
What is AI Chat for?
A homeopathy assistant that answers from the classical library first — Boericke, the repertory and the Organon — and labels anything it adds from general knowledge. It can find rubrics and add them to your case, take a case step by step, or give a quick shortlist. It stays on homeopathy.
Will it give a potency or a diagnosis?
No. Neither assistant suggests a potency or names a diagnosis, and a case with fewer than three rubrics is called too thin to rank. A clinical disclaimer sits on screen and on every printout.
More questions? See the full FAQ
Check the working, then decide
Start free — no card required. Bring a case you already know and see whether the arithmetic agrees.