Homeo AI vs RemedionAI: Cosine Search or Review Copilot?

Quick answer

Choose RemedionAI when a practitioner wants to search four classical Materia Medica corpora with natural-language symptoms, receive a cosine-similarity remedy ranking, and work within published daily analysis allowances. Choose Homeo AI when the central job is to preserve a completed case narrative, inspect proposed symptom and rubric structure, open supporting passages, correct the differential, and keep the accepted reasoning available for follow-up.

RemedionAI is unusually current. Its company announced the platform in July 2026, its public product was reviewed on 30 August 2026, and its founders link the product to a paper published in the May 2026 issue of the journal Homeopathy. That combination makes it a useful buyer comparison, but the evidence layers must remain separate. The product page describes the live offer. The paper establishes that members of the founding team published a cosine-similarity method. Neither source independently proves that the deployed product selects clinically appropriate remedies, improves patient outcomes, or is safe to use without practitioner verification.

What RemedionAI currently offers

The official page says the platform searches Allen's Keynotes, Boericke's Materia Medica, Hering's Guiding Symptoms, and Clarke's Materia Medica. A practitioner enters symptoms in natural language. The product normalises and embeds that text, compares it with the available corpora, and returns a scored shortlist with linked source material. It also advertises patient records, follow-ups, PDF reports, browser access, and account use on several devices.

Those capabilities are vendor claims, not independent performance evidence. Cosine similarity measures alignment between text representations. A higher score can be useful for retrieval, but it does not show that a symptom was interpreted correctly, that an important modality was preserved, or that the leading remedy is clinically appropriate. A ranked list can be mathematically consistent with incomplete or poorly structured input.

The pricing page is part of the same official surface. It displays a limited-period free founder offer with up to 30 analyses daily. Premium is shown at INR 1,999 monthly or INR 19,999 yearly, with up to 100 analyses daily and 25 MB of storage per user. Advanced is shown at INR 3,999 monthly or INR 39,999 yearly, with up to 500 analyses daily, multi-practitioner access, and 50 MB per user. The page also displays higher crossed-out reference prices. Treat the lower numbers as a dated founder offer, not a permanent list price. Confirm taxes, renewal, eligibility, storage measurement, quota resets, failed-request treatment, and refund terms at checkout.

What the paper establishes

The linked journal record is titled “Quantifying Symptom Similarity in Homeopathy Using Cosine Similarity Metrics.” PubMed records it in volume 115, issue 2, pages 119 to 120, with a May 2026 publication date. Several listed authors are also presented as RemedionAI founders on the product page.

That makes the paper a primary source for the method its authors studied. It is not independent validation of the company or the live software. A practitioner should ask how the published experiment maps to the production pipeline: which text editions were used, how symptoms were cleaned, which embedding model generated the vectors, how thresholds were selected, how ties and negation are handled, and whether the live implementation has been tested against a fixed set of practitioner-reviewed cases.

The product page says the platform was built on the research and mentions a patent application. A publication and a patent application show that a method has been documented or claimed. They do not establish therapeutic efficacy, live ranking accuracy, reproducibility of the deployed model, or an improvement over a repertory-led review. The correct purchasing use of the paper is to generate test questions.

Where RemedionAI may fit better

RemedionAI may fit a practitioner who wants to start in natural clinical language and search several Materia Medica corpora without first translating every expression into repertory rubrics. Its scored shortlist can help a reader decide which remedy texts deserve closer inspection. The free founder offer also creates a low-cost way to test retrieval quality before selecting a paid allowance.

The product may be especially useful for literature exploration when the practitioner already has a disciplined way to preserve the original sentence, identify missing context, and reject an attractive but incomplete semantic match. PDF export and multi-device browser access can also be practical if the resulting report preserves source identity and does not flatten corrections into a final score.

Homeo AI should not be credited with RemedionAI's exact cosine engine, four-corpus plan structure, daily quotas, or founder paper. RemedionAI should not be assumed to reproduce Homeo AI's exact narrative-to-symptom-to-rubric correction trail merely because it links sources and ranks remedies.

Where the fit needs closer testing

The reviewed RemedionAI public page makes general claims about encrypted connections, industry-standard security, authorised access, and regular backups. Its public navigation and footer did not expose separately reviewable privacy, terms, retention, deletion, subprocessor, or data-processing documents. That is a procurement gap, not proof that internal controls are absent.

Before using identifiable records, ask the contracting entity to provide dated documents that name the controller or processor roles, hosting regions, AI and cloud providers, prompt and file retention, training rules, staff access, backups, export, deletion, breach notice, account termination, and governing law. Confirm whether the small per-user storage allowances include uploaded documents, generated reports, backups, and deleted files waiting for purge.

The method also needs input-level testing. Natural-language similarity can overvalue repeated common words, miss negation, collapse laterality, and underweight timing or causation. Ask whether the interface keeps the original wording next to every matched passage, whether a practitioner can reject individual matches, and whether changing one material fact produces an understandable score change.

How Homeo AI approaches the same job

Homeo AI centres a narrower review chain. The practitioner begins with a completed narrative, checks how phrases were structured into symptoms, inspects or corrects proposed rubric mappings, opens cited passages, compares remedy differences, and preserves the accepted reasoning for a later visit. Its useful output is not a single score. It is a reconstructable chain another practitioner can challenge.

That workflow can be slower than opening a semantically ranked reading list when the only task is literature retrieval. It can be more useful when the clinic needs to explain why a phrase changed the totality, what was rejected, and how follow-up evidence altered the earlier differential. The correct comparison is total correction and review time, not the time to the first shortlist.

A reproducible clinic test

Prepare eight completed, de-identified cases already reviewed by the same practitioner. Include two clear cases, two close differentials, one sparse narrative, one case containing negation, one bilingual or regional phrase, and one follow-up. Keep an answer packet outside both systems containing the original wording, accepted symptom structure, source passages, rejected interpretations, final differential, and later changes.

In RemedionAI, record the exact input, corpora searched, matched passages, score order, time, daily allowance consumed, corrections, and export. Repeat three cases after changing one detail: a negation, a modality, or a timeline. A useful semantic system should show why the ranking moved and should not hide the original expression.

In Homeo AI, enter the same narratives. Record proposed symptoms and rubrics accepted, edited, or rejected, every passage opened, how the differential changes after correction, and whether another practitioner can reconstruct the accepted review. Reopen the follow-up and confirm the earlier rationale remains intact.

Score both products on narrative fidelity, source identity, correction time, repeatability, uncertainty, export completeness, account governance, and total cost at the clinic's real volume. Do not award either product points only because its leading remedy matches a historical prescription. Agreement can occur for the wrong reason.

Known limits and clinical responsibility

A cosine score is not a probability of benefit. A peer-reviewed method paper is not independent validation of a commercial product. Four named corpora do not prove edition accuracy, complete coverage, correct parsing, or better outcomes. RemedionAI's strengths remain worth testing, but every material claim should stay attached to its evidence category.

Homeo AI can also misread a narrative, propose an unsuitable rubric, omit relevant evidence, or produce a plausible differential that requires correction. Inspectability improves reviewability; it does not guarantee a correct conclusion.

Neither product replaces diagnosis, referral, an emergency-care decision, practitioner judgment, prescription, potency, dosage, or follow-up responsibility. Urgent, severe, or rapidly worsening symptoms require appropriate medical assessment. The practitioner remains responsible for the original record, each accepted interpretation and source, and every clinical action.

Conditional verdict

Choose RemedionAI if the clinic wants current semantic search across its named Materia Medica corpora, can verify the score and source trail with known cases, and receives satisfactory written answers about data processing, storage, deletion, commercial terms, and the production implementation of the published method.

Choose Homeo AI if the priority is a reconstructable review path from narrative through corrected symptom and rubric structure to inspected passages, remedy differentiation, and follow-up. If both are tested, preserve the same answer packet and compare measurable correction burden and governance rather than the apparent precision of a percentage or rank.

Official sources checked

• RemedionAI official product and pricing page
• RemedionAI official July 2026 launch post
• May 2026 cosine-similarity paper record

Product facts reviewed 2026-08-30. Confirm current features, prices, source libraries, and terms with the vendor.