These JARIA Terms govern the use of JARIA by End Users. They are a Standard-Form Document incorporated into the VTC End-Customer Terms (Exhibit B §1.4) and the End Customer’s Principal Agreement. They apply with Exhibit B, the Acceptable Use Policy and its Annex FU, the Privacy Policy and the Data Processing Agreement (Exhibit C). Section 12 sets out the principles ORamaVR applies to all its AI features and replaces the Responsible AI Policy (OVR-POL-013); a reference in any ORamaVR document to that policy is a reference to Section 12.
On conflict: Exhibit C prevails on personal data, except that Annex J (the JARIA Addendum) prevails on AI-specific data-protection and EU AI Act matters within its scope; Exhibit B prevails on platform-wide matters; these JARIA Terms prevail on JARIA-specific matters.
• “JARIA” means ORamaVR’s AI virtual assistant integrated into the VTC Platform, in its Base and JARIA+ configurations.
• “JARIA Base” means the JARIA functionality included with every paid VTC subscription and the Educational tier (§3.1).
• “JARIA+” means the enhanced configuration with an expanded IFU and reference corpus and the functions in §3.2, where the Order Form provides for it (not at OEM Starter).
• “End User” means a person using JARIA as a trainee, supervisor or administrator of the End Customer’s tenant.
• “Prompt” means any input an End User gives JARIA, spoken or written; “Output” means any content JARIA generates in response.
• “JARIA Addendum” means Annex J to these JARIA Terms.
Other capitalised terms have the meanings in Exhibit B and the Principal Agreement.
2.1 Acceptance. The End User accepts these JARIA Terms by click-through at first use of JARIA; JARIA is not available until then. The acceptance is recorded with the user, date and version.
2.2 Entitlement. JARIA Base comes with every paid VTC subscription and the Educational tier. JARIA+ applies where the Order Form says so. JARIA is not licensed separately. The Educational rules are in the Acceptable Use Policy, Annex FU §FU.6.
2.3 Eligibility. JARIA is for professional, faculty and student users in VTC training programmes who are at least 18 (or the age of majority where higher) and authorised by the End Customer.
3.1 JARIA Base. In-session guidance: explanations of training content, procedural steps and reference material, and answers based on the session context and the End Customer’s IFU and protocol corpus. JARIA works in real time during the session only. Usage limits, if any, are those in the Acceptable Use Policy, Annex FU.
3.2 JARIA+. JARIA Base plus session-aware feedback, adaptive guidance, and session-level information for supervisors. Post-session reports, analytics and exports are VTC Platform features, not JARIA features.
3.3 Models and limits. JARIA and OMEN image generation run through the AI providers listed in the Sub-Processor Register (OVR-POL-008). OMEN text features use the customer’s own AI account and key (“bring your own key”); that provider is not ORamaVR’s sub-processor. OMEN image generation is capped as set out in the Acceptable Use Policy, Annex FU §FU.4.
3.4 Changes to features. ORamaVR may add, change or retire JARIA features under the amendment clause of the Principal Agreement or, for an End Customer that has signed none, Exhibit B §12, with notice of material reductions.
4.1 Outputs are informational. Outputs are training aids. They may be inaccurate or incomplete and are not medical advice, a diagnosis, a treatment plan or a substitute for professional judgment. The End User is responsible for decisions taken on the basis of an Output.
4.2 Human oversight. JARIA supports decisions; it does not make them. It does not grade, certify, pass, fail or sanction anyone and makes no automated decision within GDPR Art. 22. The End Customer keeps meaningful human review wherever JARIA output informs a decision about a person, consistent with its professional, educational and accreditation standards.
4.3 Transparency. JARIA identifies itself as an AI assistant at the point of interaction: “You are talking to an AI assistant. Your spoken questions are transcribed and answered by AI. Do not mention real patients.”
4.4 Not permitted. Do not use JARIA (a) for the diagnosis or treatment of a real patient; (b) instead of an independent clinical or professional opinion; (c) against professional, ethical or institutional duties; or (d) for anything the Acceptable Use Policy or Exhibit B §4.2 prohibits. No Prompt may contain patient-identifiable data.
5.1 What is processed. Prompts (including voice audio and transcripts), Outputs, session context (scene state, IFU references), interaction metadata, the End User’s account and role, and usage counts.
5.2 Roles. ORamaVR processes JARIA data for the End Customer as its processor under Exhibit C §5. ORamaVR acts as an independent controller only for the purposes listed in Exhibit C §1.2, as described in its Privacy Policy.
5.3 No training. ORamaVR does not use Prompts, Outputs or other customer content to train or fine-tune any AI model, and its AI providers are contractually barred from doing so. Training or fine-tuning on customer content is possible only on the End Customer’s express written opt-in under this §5.3, which is the only consent path in the ORamaVR suite (Exhibit B §11.3A and the reserved Annex L-4 of the OEM Master Agreement operate only through it). ORamaVR may use aggregated, de-identified usage statistics to run and improve its services.
5.4 Location and providers. Hosting, providers and transfers are as set out in Exhibit C §7 and the Sub-Processor Register, which are not restated here.
6.1 JARIA. JARIA and all its components are ORamaVR IP (Exhibit B §7.1).
6.2 Outputs. The End Customer may use Outputs for training within its organisation. It may not use Outputs to build a competing product, to train or evaluate another AI system, or present them as written or validated by a qualified professional where others may rely on them.
6.3 No non-infringement warranty. ORamaVR gives no warranty that Outputs are free of third-party rights; apply professional judgment before reusing them.
JARIA is provided “as is” and “as available”, without warranty of accuracy, completeness, reliability or fitness for a particular purpose. ORamaVR is not liable for decisions taken in reliance on an Output, except for gross negligence or wilful misconduct or where liability cannot be excluded by law. Otherwise, liability is as limited in the Principal Agreement and Exhibit B §9. Nothing here limits liability that mandatory law does not allow to be limited.
8.1 Term. These JARIA Terms apply from acceptance while the End User uses JARIA; the entitlement ends with the VTC subscription.
8.2 Suspension. ORamaVR may suspend JARIA for a user where it reasonably suspects misuse, a patient-data Prompt or a security incident, where the VTC licence is suspended, or where a court or authority requires it.
8.3 Termination. ORamaVR may end a user’s access with notice where a ground in §8.2 persists or a material breach is not cured within the cure period in Exhibit B §11.2(a).
8.4 Survival. Sections 4, 5.2, 5.3, 6, 7, 9 and 10 survive.
ORamaVR may amend these JARIA Terms under the amendment clause of the Principal Agreement or, for an End Customer that has signed none, Exhibit B §12: 30 days’ notice for material changes, 10 days for others. Users may be asked to accept again after a material change.
Swiss law, subject to mandatory EU consumer, data-protection and AI law. Disputes follow Exhibit B §13.2 or, where the End Customer has itself signed a Principal Agreement with ORamaVR, the dispute-resolution clause of that agreement.
The End Customer is responsible for its End Users (Exhibit B §4.4). Nothing here is professional, clinical or legal advice. If a provision is invalid, the rest stands. The English text prevails.
These principles apply to JARIA, OMEN and any other AI feature of ORamaVR products, and bind ORamaVR towards its customers.
• Human oversight. AI supports training; people decide. No ORamaVR AI feature makes a decision with legal or similar effect on a trainee.
• Transparency. Users are told when they interact with AI, and AI-generated content is marked as the EU AI Act requires (Art. 50).
• No training on customer data. Customer content is never used for training without an opt-in under §5.3.
• Not medical. No ORamaVR AI feature is a medical device or gives medical advice (Regulatory Positioning Memo, OVR-REG-001). Real patient data is excluded.
• One place for limits. Usage caps are stated only in the Acceptable Use Policy, Annex FU.
• EU AI Act. ORamaVR’s working position is that its AI features, used as training tools under human oversight, are not high-risk AI systems under Annex III of the EU AI Act (Regulation (EU) 2024/1689). This position is under external review; if the review concludes otherwise, ORamaVR will meet the high-risk requirements before they apply to it. ORamaVR keeps technical documentation for JARIA on the Annex IV headings and reassesses the position on any material feature change or new guidance.
• Reporting. Report harmful or misleading AI output to security@oramavr.com (security) or support@oramavr.com (other concerns). ORamaVR investigates, corrects or withdraws features where needed, and makes any report the law requires.
• Review. ORamaVR reviews these principles at least once a year and on any material change to its AI features.
J.0 Status. Annex J carries the AI-specific contractual terms. A reference in any ORamaVR document to the “JARIA Addendum” is a reference to this Annex. Within its scope it prevails on AI-specific data-protection and EU AI Act matters.
J.1 Intended use. JARIA is a training aid only. It is not a medical device and must not be used for diagnosis, treatment, triage or any clinical decision about a real patient. JARIA+ is the expanded-corpus configuration.
J.2 Human oversight. JARIA outputs are guidance to learners and supervisors, subject to human oversight. JARIA does not grade or evaluate learning outcomes and makes no automated decision with legal or clinical effect on anyone.
J.3 Grounding and transparency. JARIA answers are grounded in the End Customer’s IFU and protocol corpus. Users are told they are interacting with AI, that content is AI-generated, and of JARIA’s known limitations and training-only scope.
J.4 Data. JARIA needs no patient data. JARIA data (voice audio, transcripts, scene state, IFU references, prompts and outputs) is processed under Exhibit C as processor data, through the providers in the Sub-Processor Register. Session and competency records are generated for the End Customer.
J.5 Responsibility and change control. ORamaVR is responsible for JARIA as set out in the Principal Agreement. Material changes to JARIA’s behaviour are version-controlled and notified under the amendment clause.