Clivora
Clivora is a Hungarian-language AI medical scribe that records a doctor-patient consultation or a dictation and turns it into a structured clinical report, with claimed accuracy above 95 percent and far less typing.
What is Clivora?
Clivora is an AI medical scribe built specifically for the Hungarian clinical language. It listens to a consultation, or to a physician dictating, and returns a structured document: a clinical report, a referral letter, or a fitness-for-work opinion. The terms of service define three usage modes. In dictation mode it only produces a transcript. In consultation mode it turns the dialogue between doctor and patient into a report. In instrument examination mode it builds the report from what the staff say while the examination is running. The system was trained on more than a hundred hours of professional audio collected with the help of medical scribes and students, using real reports and real consultations rather than generic speech data. It was also designed for the acoustics of an actual consulting room, where machines beep and colleagues talk over each other. The vendor publishes a set of claimed figures: accuracy above 95 percent, up to 75 percent less time spent writing reports, consultations shortened by 15 percent, five to ten minutes saved per patient, and more than a hundred healthcare professionals using it. Templates are central to the product. Ready-made schemas exist per specialty and per Hungarian medical system, and users can build and save their own; Enterprise customers get institution-wide templates so that everyone documents the same way. Integration with the main Hungarian medical systems is offered on request, and in an integrated setup Clivora can see prior patient data, which the vendor says makes the reports more accurate. The responsibility model is stated plainly and repeatedly. Clivora never decides anything: every draft is checked, corrected and approved by the treating physician, and the contract explicitly excludes using the tool for diagnosis, prevention, monitoring, prognosis or treatment. The vendor also states that the solution is not an accessory to a medical device. It is delivered as a web application, with an option to run the whole containerised environment on the institution's own infrastructure.
What it does
- Transcribe a doctor-patient consultation in Hungarian in real time
- Turn that transcript into a structured clinical report following a chosen template
- Handle plain dictation, producing a transcript without structuring it
- Generate a report from staff speech recorded during an instrument-based examination
- Produce referral letters and fitness-for-work opinions alongside standard reports
- Let the clinician review, edit and refine the generated document before approving it
- Send a finished report to another device through an optional encrypted transfer
When to use Clivora / When not to
A quick filter to help you decide if Clivora is the right fit.
When to use Clivora
- Hungarian physicians running outpatient consultations, in the specialties the site names itself: family medicine, cardiology, dermatology, orthopaedics, gastroenterology and radiology
- Group practices and clinics that want every clinician working from the same institution-wide templates, which is what the Team and Enterprise tiers are built around
- Practice assistants, nurses and medical scribes who carry the typing burden today and are named as beneficiaries in the vendor's own blog
- Hospital IT managers who need the tool to sit beside an existing medical system rather than replace it, since integration is offered on request
- Institutions with sovereignty constraints, because the whole containerised environment can be run on the institution's own infrastructure under a specific agreement
When not to use Clivora
- Anyone outside Hungary or working in another language: the interface, the models, the contract and the prices are Hungarian only
- Individual consumers, since the terms restrict ordering to non-consumer entities that qualify as healthcare providers under Hungarian decree 96/2003
- Clinicians looking for diagnostic or therapeutic support, which the contract explicitly excludes through its 'excluded scope of application' clause
- Teams that need a mobile app or a public API, neither of which the site documents anywhere
- Practices unwilling to buy a decent external microphone, because the vendor repeatedly states that built-in microphones cannot handle clinic noise
How to use Clivora
A typical end-to-end flow, from setup to results.
- Create an account or sign in at the web application, using an email and password or a Google login
- Start on the permanently free tier, limited to five transcriptions, to see how the output reads
- Get a good external microphone: the vendor states plainly that built-in microphones cannot filter clinic noise
- Ask the team for help choosing and positioning the microphone, which they offer as part of onboarding
- Pick the template that matches your specialty and your medical system, or build your own
- Choose the mode: dictation, consultation, or instrument examination
- Start recording at the beginning of the encounter and speak naturally near the microphone
- Receive the draft report, which the vendor commits to returning within one minute
- Review, correct and refine the text, then attach it to the correct patient and encounter
- Approve the report and file it in your institution's medical system
Pros & Cons
Pros
- Genuinely built for Hungarian medical language, so no foreign-language skills are needed to use it
- Time savings are quantified rather than vague: up to 75 percent less writing, five to ten minutes per patient
- The retention model is strong: audio and transcript are never stored and disappear within ten minutes at most
- Processing sits in the EU by default, in Germany, Sweden and Ireland, with an on-premise option
- A permanently free tier lets a clinician test the output quality before committing anything
- Setup is genuinely light: a few hours, no technical skills, with personal training included
- Templates adapt to the existing medical system, so institutions do not have to replace their software
Cons
- Hungarian only, with no other interface or processing language announced anywhere
- The service is provided 'as is', with no warranty, no fitness guarantee and no promised outcome
- The vendor disclaims responsibility for the accuracy and completeness of generated text; the whole verification burden sits with the physician
- Templates and schemas created by the customer transfer irrevocably to the vendor, who may open them to third parties
- No mobile app, no documented public API and no social presence to check the company against
- Output quality depends heavily on microphone quality and ambient noise, which the vendor acknowledges
- The named subprocessor list for the AI system is only released to the data protection officers of contracted institutions
Pricing & Plans
A permanently free plan is available at 0 HUF per month, limited to five AI transcriptions. The cheapest paid entry point is the Starter plan at 40,000 HUF per user per month, quoted net of tax. Annexe 2 of the terms confirms the same figure as the per-user monthly service fee.
- intuitive interface
- five AI-driven Hungarian transcriptions with basic features
- a personal template
- 10+ supported medical specialties
- email administrator support
- personal training for one user
- unlimited Hungarian AI transcription with basic features
- unlimited personal templates
- 10+ specialties
- email administrator support
- personal training for five users
- unlimited transcription
- unlimited personal templates
- 20+ specialties
- email and phone administrator support
- group training
- unlimited transcription with pro features
- unlimited institution-wide templates
- 50+ specialties
- email and phone administrator support
- Annual billing carries two months of discount plus a free professional microphone
- the terms also allow discounts for annual contracts
- several users
- or individual assessment
Data, GDPR & hosting
A consolidated view of how Clivora handles your data.
GDPR overview
Implementation is unusually concrete for a product this young. Two separate notices exist: a general one for the site, accounts, newsletter, cookies and invoicing, version 2.0 effective 10 July 2026, and a dedicated notice for the AI system. Roles are stated explicitly, the institution being the controller and Clivora the Article 28 processor. A data protection officer is named, Veres Gábor, with published contact details. Legal bases are cited article by article, including Article 9(2)(h) for health data and Hungarian act XLVII of 1997. Article 22 is addressed directly: no solely automated decision-making, since every draft is checked and approved by the treating physician. The one non-EU transfer relies on the European Commission's standard contractual clauses. The supervisory authority named is the Hungarian NAIH.
Who owns the data?
Two very different regimes sit side by side. For patient material, Clivora acts as a processor under Article 28 GDPR: the healthcare institution is the controller, the audio and transcript are handled only on its documented instructions, and the approved report stays in the institution's own system. For templates and schemas, the terms are far more one-sided. Clause 6.3.1 states that any template or schema created by the customer or its staff transfers automatically and irrevocably to the vendor, moral rights are waived, and the vendor may use, modify, develop and make them available to third parties. Source code and know-how are never transferred to the customer.
Reuse rights
The approved report belongs to the institution's medical documentation and is retained there under Hungarian health-records law, so the clinician can reuse it freely as part of the patient file. Everything else is narrower. The access right granted by the contract is non-exclusive and non-transferable, the underlying software, source code and know-how are never handed over, and site content may not be copied, downloaded, redistributed or resold without prior written consent. The customer must have the draft checked by the staff who attended the consultation and must make every correction itself. After termination there is a thirty-day window to download archived data, after which the vendor deletes it. Templates created by the customer become the vendor's property and may be opened to other users.
Data retention & training
Hosting summary
Application infrastructure runs by default in the European Union, in Germany, at Frankfurt. Speech recognition and report structuring each come in two variants chosen by the institution: an in-house model, run either on the institution's own infrastructure or on dedicated EU endpoints in Ireland, or a cloud model processing exclusively in Sweden, which is the default. Under a specific agreement the entire containerised environment, with the institution's own database, can be hosted on site, in which case nothing leaves the building. One component may sit outside the EU: an optional language-rephrasing step handled by a United States AI provider. It is enabled by default but can be switched off per institution, receives only pseudonymised data with direct identifiers stripped, and is technically prevented from ever receiving the raw transcript. That transfer relies on the European Commission's standard contractual clauses and a processing agreement, and the provider does not use the data for model training. Note that the domain's IP address geolocates to the United States, but that is a Cloudflare anycast node, not the application host.
Where Clivora works
Country-level availability.
Available in
Things to keep in mind
Risks and trade-offs to weigh before adopting Clivora.
- The tool captures special-category health data by voice, so any misconfiguration hurts patients who are not the customer and never chose the product
- The vendor itself warns that pseudonymisation is not anonymisation: a name spoken mid-sentence can slip past the filter before data leaves the EU
- The non-EU rephrasing step is active in the default configuration, so an institution must take deliberate action to switch it off
- The terms warn that the system may still produce diagnostic or therapeutic suggestions even when instructed not to, and the physician must spot and delete them
- When the report writes itself, careful proofreading is the first thing a rushed clinician drops, which is exactly where errors become permanent
- Long-term reliance may erode the clinical writing skill that lets a doctor notice what a summary has quietly left out
- Templates and schemas built by the user contractually stop being the user's, and may be made available to other customers
Setup & Integrations
Technical difficulty
Low. The vendor states that no technical knowledge is needed, the interface is intuitive, and a practice can be running within a few hours. You create an account with an email and password or a Google login. The real constraint is hardware, not software: a quality external microphone is effectively mandatory, and the team helps choose and position it. Personal training is included, followed by email support, with phone support from the Team tier upwards. Medical-system integration and on-premise deployment are heavier projects, quoted individually, and the customer keeps responsibility for its own network and firewall setup.
Deployment
Supported languages
Behind Clivora
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What exactly does Clivora do?
How accurate is the transcription?
Is there a free version or a demo?
Which medical specialties are supported?
Does it work with our existing medical system?
Are patient recordings stored anywhere?
Where is the data processed?
Is Clivora GDPR compliant?
Does the AI make clinical decisions?
Who is allowed to subscribe?
Should you pick Clivora?
Clivora is a deliberately narrow product: one language, one profession, one country. That narrowness is its strongest argument. A general-purpose scribe transcribing Hungarian clinical speech would struggle with the terminology, and the vendor's decision to train on real reports and real consultations, collected inside a working gastroenterology chain, is a more convincing origin story than most. The legal documentation is also unusually thorough for a company this young. Two separate privacy notices, processing locations named component by component, a template patient consent form handed to institutions, and a retention model where nothing is kept at all beyond a ten-minute technical window: these are not the artefacts of a team improvising its compliance. The counterweight sits in the contract. The service is supplied as is, the vendor disclaims responsibility for the accuracy of what its own model writes, and the entire verification burden lands on the physician who signs. Clause 6.3.1 goes further and hands the vendor irrevocable ownership of any template a customer builds, with the right to open it to other users. Buyers should read that clause before investing effort in custom schemas. The performance numbers deserve the same caution. Accuracy above 95 percent, seventy-five percent less writing time, consultations fifteen percent shorter, more than a hundred physicians on board: all are self-declared, with no published study or independent source behind them. The company is young too, with a domain registered in May 2025 and no Wayback history at all. For a Hungarian practice drowning in administration, the free tier makes the question cheap to answer empirically. For anyone outside Hungary, it is simply not available.
- Choosing a selection results in a full page refresh.
- Opens in a new window.