Perceptors
Perceptors is an AI native learning platform for medical societies, universities, health systems and healthcare employers that want to turn their own approved clinical expertise into certified courses. A program team supplies a vetted set of sources, an audience and learning goals. Evidence, Curriculum, Assessment, Localization and Quality agents then map the sources, draft modules and assessments with answer rationales, translate material and flag unsupported statements, and four human approval gates, from source map to release, mean nothing publishes on its own. Every release records its sources, drafts and reviewers.
Learners use a Guide that answers only from the approved course materials and shows citations, declining or escalating questions outside them, and learning signals can propose changes without rewriting published content. Administrators see progress and assessment results across a cohort, and institutions can run a white label or co branded academy alongside an existing learning management system, issuing certificates under the rules of the named issuer. Forward deployed engineers set up an academy in about 30 days, and learner interactions exclude patient information.
The GCLS AI Lab at the Geneva College of Longevity Science operates Perceptors, with DrVibe Inc. as the contracting party, and its cofounders include Fady Hannah-Shmouni, Omar Saleem and Dominik Thor. Live programs include the GCLS AI Academy and Hormonaly Academy, and partner logos include Hamad Medical Corporation and the University of British Columbia. A free institutional pilot covers one course and 100 learners, and Academy plans start at $99 per learner.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Agents do the production work. Specialized agents map an approved evidence collection, draft modules, build assessments with answer rationales, translate material and check quality, and the learner Guide answers questions from the selected course sources with citations. Those workflows are why an education team would choose this over a conventional learning platform, and a Clinical AI Studio and a books imprint apply the same agents to other formats.
Course hosting, progress tracking, certificates and the review desk are ordinary learning management functions around that core. Faculty stay responsible for what is taught and approve every release, including courses delivered through a white label academy.
Each production stage has a named approver: a subject lead for the source map, a program editor for the draft, a clinical reviewer for the quality check and the program owner for release. Any of them can revise, reject or return work before learners see it, and the platform states that nothing publishes on its own.
The Guide works differently. It runs live during a course from an approved source collection, declining or escalating questions outside it, and its individual replies do not wait for human approval. Course publication is gated while real time learning support works within the boundary those gates set.
The technology page describes an agent workflow covering evidence retrieval, curriculum, assessment, translation, delivery and measurement, carried out by five named agents, Evidence, Curriculum, Assessment, Localization and Quality, that share context under defined permissions. The Guide uses generation grounded in the course sources and cites its evidence, the Quality agent flags unsupported draft statements for review, and approved changes are logged. Each program chooses the evidence its course is grounded in.
The inputs, drafts and releases are explained. Model architecture by version, training data and full quality measures for the Guide and the separate agents are not given.
OpenRouter, ElevenLabs and Anthropic's Claude are listed among the tools and organizations supporting the work, and the privacy policy says AI providers may process what is needed for specific features. That points to a routing layer across several model providers plus a voice provider.
Which model and provider serve each course, Guide or assessment agent in production is not mapped, and neither versions, deployment regions nor any right to choose a model are given. The academy provider selects and approves the course sources, which are separate from the data the language models were trained on.
Live programs are the evidence. The GCLS AI Academy launched in December 2025, Hormonaly Academy opened with a free skin peptides course for clinicians in November 2026, and partner logos include Hamad Medical Corporation, the University of British Columbia, Sigmund Freud Private University and Universitas Muhammadiyah Jakarta. Hormonaly's launch release says the platform is used by more than 30 academic centers and on track for 100,000 learners in 2027.
Hormonaly is affiliated with a cofounder, and the company itself notes that an association does not prove a paid deployment. No completed evaluation reports learner counts, comparison groups or measured gains in competency or practice.
The privacy policy forbids entering patient information, sensitive health details or confidential information into the AI learning features. Questions, responses and course context are processed through service providers, optional remembered preferences can be controlled by the learner, and only the relevant course administrators and faculty can see learning records. The Guide answers only from approved materials.
Retention varies by academy and record type. No deletion timetable is given for Guide prompts, and there is no general commitment that learner questions or course materials stay out of model improvement with every AI provider used.
Users are told to keep protected patient information out of learner records and Guide conversations. The privacy policy defines the account, learning and assessment records involved and says certain AI providers process what the Guide needs, and the trust page says HIPAA requirements can be discussed and scoped with institutions.
No business associate agreement is published and no plan permits entering identifiable records, so the patient data limits apply to ordinary academies even when an institution explores more. The standard academy works from approved teaching evidence and educational records rather than patient data.
The trust page describes bounded agent permissions, approval roles, versioned sources and a change record, and the privacy policy adds access control and operational safeguards for learner information, with DrVibe Inc., the Delaware contracting entity, as the data controller responsible for them. Perceptors says it supports SOC 2 and HITRUST requirements as needed, scoped during discovery.
That statement names no completed audit, certified entity or current attestation for the platform, and no report covers the Guide or the AI providers behind it. Governance and hosting questions are handled in an institution's evaluation or plan negotiation.
The learner terms describe the platform and its AI responses as educational and leave medical judgment and evaluation of individual patients to qualified professionals. The system drafts course content and answers within an approved curriculum, with no function for assessing a patient or suggesting a treatment.
Certificates depend on the course and its named issuer, such as the Geneva College of Longevity Science for Hormonaly graduates, and academic or professional credit applies only where explicitly confirmed. No device clearance or exclusion analysis is published, and references to clinical practice in courses describe the subject being taught.
Content governance is spelled out. The trust documentation names reviewers, permitted sources, release decisions and version history for each academy, a Quality agent flags unsupported material, and an editor or clinical reviewer can reject a draft. Translations go through editorial review before release, learning signals can only propose changes rather than rewrite published content, and institutions can test the Guide against their own source collection.
No results show how the Guide performs by language, learner population or accessibility need. Review records let an institution investigate a particular release, but consistency across learner groups has not been measured.
Reviewers can send a draft back, change a source or stop a planned release, and the approval record traces every published version to its inputs and reviewer. Because learning signals can only propose changes, a fix to a live course passes through the same review as the original. The learner terms warn that Guide responses may contain errors and separate the course creator's responsibilities from the platform's, and support handles access, certificate and billing requests.
No remedy is set out for an incorrect Guide answer, no educational accuracy is guaranteed, and no time limit applies to correcting an error learners can see in a live course.
Each academy is a standalone branded site that can coexist with an institution's existing learning management system, in white label or co branded form so the institution keeps its own identity, and custom integrations are scoped on institutional plans. Enrollments, assessments and progress are the service's own records, and a public certificate check confirms a learner's credential.
No clinical record integration is offered, and none would be expected, since learner records are meant to exclude identifiable patient data. No named learning platform has a documented two way connection either, so integration terms are worked out in discovery.
Perceptors hosts each academy, which can run alongside an institution's existing learning tools, and the platform is now live in the United Arab Emirates with GCLS courses. The privacy policy says processing locations depend on academy hosting and the providers involved and can span more than one country, and institutional data requirements are agreed during discovery.
A Delaware contracting entity and general contact locations do not establish a data region. No customer hosted option or selectable region is offered, and no separate commitments cover learner, assessment, Guide and certificate data or their backups.
The free pilot includes one branded course, analytics and up to 100 learners. Academy starts at $99 per learner, counting anyone enrolled in at least one course during the term, and the institutional tier is quoted and can add integrations and support. Standard academies carry no setup fee, while custom builds are priced separately, and forward deployed engineers handle a launch of about 30 days.
Pilot and institutional requests go through a briefing form. Individual course prices and enrollment, such as Hormonaly's paid certifications, are set by each academy under its own terms, separate from buying the platform.
Academies can host professional education for physicians, nurses, pharmacists, clinical leaders and other healthcare staff, as well as custom workforce training. Live programs cover clinical AI and longevity medicine at GCLS and peptide and hormone certification at Hormonaly, and a longevity nutrition academy is listed as launching. Partner institutions run from Qatar's Hamad Medical Corporation to universities in Canada, Austria and Indonesia.
The nursing use case notes that a nursing program has not yet been published. Named live offerings show use in selected settings, but completed deployments across every specialty on the use case pages, and accreditation of those programs, have not been shown.
Alternatives to Perceptors
Other products that do the same kind of work as Perceptors, and what each one does.
Xuron provides AI virtual humans for practicing healthcare conversations. Learners speak with simulated patients, caregivers or clinicians, receive feedback and repeat an activity.
MedVR Education provides virtual patient encounters and XR clinical skills training for universities, nursing programs and hospital educators.
Oscer gives medical students and other trainees virtual patients for practicing objective structured clinical examinations.
C-SATS (Crowd Sourced Assessment of Technical Skills) is a surgical video assessment platform owned by Johnson and Johnson MedTech.
SimConverse provides virtual patients and assessment tools for healthcare communication education. Learners move through preparation, spoken conversation, documentation, feedback and reflection.
AI clinical simulation for nursing education and health system workforce training, delivered in immersive VR and screen-based formats.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Academy from $99 per learner
$99 baseline
|
The pilot covers one course and up to 100 learners. Academy starts at $99 per learner enrolled in at least one course during the term. Institutional volume is quoted. | Learning interactions exclude patient information. Institutional HIPAA requirements can be scoped during discovery, but the company does not publish a business associate agreement or tier. | Standard academies have no setup fee, and forward deployed engineers handle a launch of about 30 days. Custom builds, integrations and additional institutional services are quoted. | Vendor Published |
Pilot and platform plans begin with a briefing request. The price and access period of individual courses are set by the academy that offers them.