Homeo AI vs hpathy.ai: Guided Case-Taking or Review Copilot?

Quick answer

hpathy.ai is worth evaluating when a qualified practitioner wants a guided consultation that asks differentiating questions and ends with one rubric-grounded recommendation. Homeo AI fits a different starting point: the practitioner already has a completed narrative and wants to inspect proposed symptom and rubric mappings, examine supporting passages, and review the differential before accepting any reasoning.

The decision is about where assistance enters the work. An interview can help expose missing information before analysis. A review copilot can help scrutinize the interpretation of information already collected. Neither route makes a recommendation clinically correct merely because a repertory or AI model was involved. Choose the workflow whose intermediate steps you can check, rather than the one with the most confident final answer.

This comparison uses first-party product, plans, terms, privacy, and attribution evidence supplied for 4 October 2026. These sources establish the vendor's published claims, not an independent product trial, clinical validation, or security audit. The Homeo AI description follows its established practitioner-review positioning; account-specific availability and current commercial terms still need confirmation.

What the guided route actually promises

hpathy.ai states that its corpus includes four classical repertories, six materia medica works, 87,853 rubrics, 42,934 sections, and 2,616 remedies. It describes up to ten differentiating questions followed by a single rubric-grounded recommendation, and states that it uses Claude. Those details give buyers something more concrete than a generic claim of AI assistance.

The possible advantage is a bounded interview. A practitioner can assess whether each new question resolves an uncertainty that matters to the case. The ten-question ceiling also creates a useful procurement question: what happens when important uncertainty remains after the permitted questions? Do not assume that reaching the end means the narrative is complete or that the system has excluded every relevant alternative.

A single recommendation can make the endpoint easy to understand, but it can also hide competing interpretations. Ask whether the practitioner can see the selected rubrics, original wording, supporting source passages, alternatives considered, and remaining uncertainty. The evidence supplied here does not establish every one of those controls. They are trial requirements, not features credited to hpathy.ai without verification.

Homeo AI's stated review workflow starts from a completed account rather than promising this exact guided interview. Its comparison value lies in making the narrative-to-symptom-to-rubric interpretation available for practitioner inspection and correction. Do not credit it with hpathy.ai's question sequence, corpus counts, or single-recommendation architecture merely because both products use AI.

Follow the question back to the source

For an interview tool, source provenance starts before the final output. A differentiating question selects what information receives attention. Ask which earlier observation triggered the question, what rubric distinction it is testing, and whether a neutral question was used. A question that assumes a missing modality can steer the answer toward a pattern that was never present.

During a trial, keep four items together: the original statement, the question asked, the answer supplied, and the source used to interpret it. Then inspect the translation. Did a statement about one circumstance become a general symptom? Was a negation preserved? Was a tentative answer treated as definite? Can the reviewer trace the final recommendation through the actual responses rather than through a polished summary alone?

For a post-case review copilot, start at the opposite end. Identify an interpretation that you would challenge and return to its supporting text. Check the source title, edition where available, rubric hierarchy, and passage context. A citation to a real book is insufficient if the cited passage does not support the mapping. Record uncertainty or disagreement rather than accepting a plausible explanation just to complete the review.

These checks measure inspectability and fidelity. They do not establish therapeutic efficacy or turn classical source material into modern clinical outcome evidence. A source-grounded mistake is still a mistake.

A hypothetical interview with one ambiguous statement

Use de-identified cases created entirely for software evaluation. Consider an invented completed narrative in which an adult reports occasional disturbed sleep and says, "I feel better outside when I walk, but sitting outside makes no difference." A later sentence says that waking happens on some nights, not every night. No patient, diagnosis, historical prescription, or treatment outcome is represented by this example.

Before entering it, write down the ambiguity: the reported change could relate to movement, outdoor conditions, or their combination. Do not choose a remedy or declare a rubric correct in advance. The reference packet should preserve the exact wording and explain what clarification would be needed to distinguish those interpretations.

In a hpathy.ai evaluation, observe whether a differentiating question distinguishes walking from fresh air without implying the answer. If the interviewer asks only about outdoor conditions, record the unresolved movement question. Preserve the supplied answer and inspect whether the final recommendation treats occasional waking as nightly waking. A failure should be documented as a mismatch between wording and interpretation, not as proof that another remedy would work.

In Homeo AI, submit the completed hypothetical narrative and inspect the proposed mappings. Reject an unsupported generalization from walking outdoors to improvement from outdoor conditions alone. Check that frequency and negation remain visible in the structured case. Record whether the supporting passage can be opened and whether the accepted rationale can be understood by another qualified reviewer.

Now change only one sentence to say that walking indoors also helps. Repeat the review and compare the reasoning, not merely the leading remedy name. The change should matter because it clarifies the observation. If either system changes its output without a reviewable explanation, or ignores the clarification, that is a useful software finding. No clinical action should follow this hypothetical exercise.

Attribution answers ownership questions, not accuracy questions

hpathy.ai's attribution page, updated on 26 September 2026, documents public-domain and OOREP GPL sources. This is useful visibility into the origin of the material and software used. It helps a buyer ask which text collections were incorporated, how they were digitized, and where source credits or license information can be inspected.

Keep three questions separate. What licenses apply to the underlying material? How faithfully does the digital corpus reproduce it? How appropriately does the application use that material in this case? Public-domain status answers an ownership question; it does not guarantee transcription accuracy, complete indexing, edition equivalence, or sound interpretation. A GPL attribution does not establish that every service component, database, or output is openly licensed.

If the clinic needs to export passages, redistribute teaching materials, or integrate the service into another application, request written clarification of the relevant permissions. For ordinary evaluation, sample a passage against a source the practitioner can inspect independently. Note omissions, changed spelling, rubric hierarchy, and edition differences. Do not infer corpus quality from the size of the published counts.

Saved records and submitted text have different boundaries

hpathy.ai's privacy evidence says submitted consultation text goes to third-party AI providers, while saved patient records are not sent to AI. The product separately states Claude use. The important boundary is the content being submitted, not whether the interface calls it a saved record. Copying information from a record into a consultation can place that information in the externally processed text.

That distinction should shape the clinic's trial procedure. Use invented narratives without names, contacts, exact dates, addresses, images, record numbers, or unusual combinations that could identify a person. Keep any real record and re-identification information outside the trial. Do not interpret the saved-record statement as a blanket promise that all information entered anywhere stays away from external providers.

Before identifiable practice use, ask which provider receives which fields, whether attachments or metadata are included, where processing occurs, how prompts and outputs are retained, whether training or evaluation use is permitted, and what deletion covers. The supplied evidence does not settle every one of these questions. Obtain the terms that govern the intended account and reconcile them with the clinic's own obligations.

Apply the same scrutiny to Homeo AI. Its review-oriented positioning is not evidence of a particular hosting location, model-provider contract, retention schedule, or no-training guarantee. Source visibility and data governance solve different problems; a buyer needs satisfactory answers to both.

Current pricing and a short procurement checklist

As of the supplied 4 October evidence, hpathy.ai's Free plan is USD 0 with five consultations per month. Pro is marked TBD and coming soon. There is no confirmed Pro price, release date, paid capacity, or service commitment in this evidence set. Do not budget a future plan as if it were currently available or assume that free access includes a production support agreement.

Five consultations can support a small workflow check, but the buyer should establish what counts toward the allowance. Ask how revisions, abandoned interviews, repeated submissions, and follow-ups are counted. These rules are unknown here. Homeo AI's current price is also not established by this comparison packet, so confirm the intended plan and usage limits rather than calculating an unsupported price advantage.

Before purchase or wider adoption, obtain answers to five practical questions:

• Can another practitioner reconstruct why each question and final rubric were selected?
• Can uncertainty, incorrect mappings, and unsupported recommendations be rejected without losing the original narrative?
• Which source editions, attribution notices, and export permissions apply to the actual account?
• Which submitted fields reach AI providers, and what retention, deletion, and access terms govern them?
• What happens when the free allowance is exhausted or the coming-soon plan changes?

Keep the answer packet with the dated plans and terms. hpathy.ai's terms were updated on 19 September 2026; that date helps identify the document reviewed but does not establish that every future account has the same terms.

Choose by the reasoning you need to inspect

Choose hpathy.ai for a controlled trial when guided clarification is the main need, the current five-consultation allowance is adequate, and the question-to-source-to-recommendation chain survives inspection. A bounded interview is useful only when the practitioner can recognize what remains unanswered.

Choose Homeo AI when the main need is to challenge and document interpretation after case-taking, using the completed narrative as the starting point. Test the actual correction and source-review experience before assuming that its positioning meets the clinic's requirements. The useful deliverable is an accepted human reasoning record, not an AI answer treated as authority.

Neither product replaces diagnosis, referral, an emergency-care decision, professional judgment, prescription, potency, dosage, or follow-up responsibility. This article gives no treatment instructions or outcome claims. If appropriate medical assessment is required, a software recommendation must not delay it. The practitioner remains responsible for every accepted interpretation and subsequent clinical decision.

Official sources checked

• hpathy.ai official product page
• hpathy.ai Free and coming-soon Pro plans
• hpathy.ai terms updated 19 September 2026
• hpathy.ai consultation and saved-record privacy
• hpathy.ai attribution updated 26 September 2026

Product facts reviewed 2026-10-04. Confirm current features, prices, source libraries, and terms with the vendor.