Myukura|Myukura Trust Center
See how it worksBeta Access

MYUKURA

Myukura is building clinical AI technology to reduce administrative burden for healthcare professionals. Its platform helps clinicians transform consultations into structured documentation, clearer patient communication, and supervised post-consultation workflows.

Myukura's mission is to make clinical work faster, simpler, and more supportive, while keeping healthcare professionals in control.

Sections

The examples below are simplified representations of Myukura's AI governance model. They are designed to explain how Myukura approaches clinical AI boundaries, data minimisation, human review and operational controls. Actual behaviour may vary depending on the selected product mode, deployment environment and customer configuration. Some capabilities are available in Myukura Core, while others are designed for Edge or Sovereign deployment scenarios.

AI Governance

Myukura is designed for supervised clinical workflows, not autonomous clinical decision-making. Our AI systems support documentation, communication and operational task management while keeping clinical judgment with the healthcare professional.

Myukura does not provide autonomous diagnosis, does not recommend treatments, and does not send patient-facing communications without professional approval.

Raw audio is not stored. Myukura keeps the validated clinical note, not the recording.

Clinical AI Boundaries

Core

The following configuration is an illustrative example of Myukura's safety-by-design approach. It does not represent the full internal system, but it shows the type of boundaries, permissions and workflow controls that guide how Myukura handles clinical AI features.

{
  "allow_diagnosis": false,
  "allow_treatment_recommendations": false,
  "allow_autonomous_clinical_decisions": false,
  "allow_auto_send": false,
  "require_human_approval": true,
  "patient_facing_outputs": "draft_or_preview_until_approved",
  "clinical_note_status": "draft_until_professional_validation",
  "raw_audio_storage": false,
  "data_minimisation": true,
  "audit_logging": true
}

These boundaries reflect a simple principle: Myukura can assist with documentation, summaries, tasks and communication drafts, but the healthcare professional remains responsible for reviewing, validating and approving outputs before use.

Data Minimisation by Design

Core

Myukura is designed to minimise the amount of data retained at each step of the workflow. Raw consultation audio is not stored as a permanent record. The system is designed to process only what is needed to generate structured documentation and workflow outputs.

The persistent asset is the validated clinical note, associated workflow metadata and audit trail — not the raw audio recording.

{
  "data_minimisation": true,
  "store_raw_audio": false,
  "persistent_record": "validated_clinical_note",
  "store_transient_processing_artifacts": false,
  "retain_only": [
    "validated_note",
    "approved_outputs",
    "workflow_metadata",
    "audit_log"
  ]
}

This approach reduces unnecessary data exposure, simplifies auditability and keeps Myukura focused on clinical documentation and operational workflow support.

Audio Processing and Deletion

Core

Myukura does not store raw audio by default. Depending on the selected processing mode, audio can be processed locally, processed temporarily, or transformed into text for clinical documentation. Once processing is complete, raw audio is discarded.

{
  "raw_audio_storage": false,
  "raw_audio_used_for": "temporary_processing_only",
  "delete_audio_after_processing": true,
  "store_validated_clinical_note": true,
  "store_raw_recording_as_medical_record": false
}

In practical terms:

  1. The consultation is captured.
  2. Audio is processed to support transcription or structured documentation.
  3. Raw audio is discarded after processing.
  4. The professional reviews and validates the generated note.
  5. Myukura stores the validated note, selected metadata and audit trail.
  6. Raw audio is not retained as the long-term clinical record.

Processing Modes

Myukura can support different processing modes depending on the clinical workflow, privacy requirements and deployment environment.

Cloud Assist Mode

Core

In Cloud Assist Mode, consultation content may be processed in the cloud to generate structured documentation, summaries or operational outputs. Raw audio is used only for temporary processing and is not stored as a permanent asset.

{
  "mode": "cloud_assist",
  "cloud_processing": true,
  "store_raw_audio": false,
  "raw_audio_used_for": "temporary_processing_only",
  "delete_audio_after_processing": true,
  "store_validated_clinical_note": true,
  "require_human_approval": true,
  "allow_auto_send": false
}

Example workflow:

  1. The consultation is captured.
  2. Audio is processed to generate a draft note.
  3. Myukura creates structured documentation and optional post-consultation drafts.
  4. Raw audio is discarded after processing.
  5. The professional reviews and validates the clinical note.
  6. The validated note, approved outputs, workflow metadata and audit trail remain available according to the clinic's configuration.

Local Privacy Mode

Edge / Sovereign

In Local Privacy Mode, audio processing is designed to happen locally on the device or controlled environment. This mode is intended for workflows where raw audio should not leave the local environment.

{
  "mode": "local_privacy",
  "process_audio_locally": true,
  "cloud_processing": false,
  "store_raw_audio": false,
  "local_transcription": true,
  "local_task_extraction": true,
  "delete_audio_after_processing": true,
  "require_human_approval": true,
  "allow_auto_send": false
}

Example workflow:

  1. The consultation is captured locally.
  2. Audio is transcribed locally.
  3. Raw audio is discarded after processing.
  4. Myukura generates a local draft or structured output.
  5. The professional reviews the result.
  6. Any external action remains blocked until approval.

Hybrid Mode

Edge / Sovereign

Hybrid Mode combines local processing with controlled cloud assistance. For example, audio can be processed locally while only a minimised transcript or structured text is sent to the cloud for higher-quality drafting.

{
  "mode": "hybrid",
  "process_audio_locally": true,
  "send_raw_audio_to_cloud": false,
  "send_minimised_text_to_cloud": true,
  "cloud_processing": "structured_text_only",
  "store_raw_audio": false,
  "delete_audio_after_processing": true,
  "require_human_approval": true,
  "allow_auto_send": false
}

Example workflow:

  1. Audio is captured and processed locally.
  2. Raw audio remains local during processing.
  3. Raw audio is discarded after processing.
  4. Only minimised text or structured content may be used for cloud-assisted drafting.
  5. Myukura generates a clinical note draft, patient-friendly summary or operational tasks.
  6. The professional reviews and approves before anything is saved as final or sent externally.

Human-in-the-loop by Default

Core

Myukura is built around professional review. AI-generated outputs remain drafts or previews until the healthcare professional validates them.

{
  "generated_soap_note": "draft",
  "patient_summary": "preview",
  "email_to_patient": "draft",
  "whatsapp_message": "preview",
  "calendar_action": "pending_approval",
  "task_creation": "reviewable",
  "finalization_requires": "professional_validation"
}

This applies to clinical notes, patient-friendly explanations, follow-up messages, task extraction and external communication workflows.

Post-consultation Agent Controls

Core

Myukura's post-consultation agents are designed to work from professional-approved or reviewable source material. They can help transform a clinical note into clearer communication, tasks or follow-up drafts, but they are not designed to create new clinical decisions.

{
  "source_required": "approved_or_reviewable_clinical_note",
  "allow_new_diagnosis": false,
  "allow_new_treatment_plan": false,
  "allow_medication_changes": false,
  "allow_dose_changes": false,
  "allow_patient_facing_send_without_approval": false,
  "compare_output_against_source": true,
  "flag_missing_or_added_plan_items": true
}

Example supported actions:

  • Generate a patient-friendly explanation from a clinical note.
  • Translate approved instructions into another language.
  • Prepare an email draft for the patient.
  • Prepare a WhatsApp preview.
  • Extract administrative tasks.
  • Create a follow-up reminder pending approval.
  • Flag unclear or sensitive content for professional review.

Safety Gate Examples

Core

Myukura applies safety checks before outputs are finalized or sent externally.

{
  "blocked_examples": [
    "new_diagnosis_not_present_in_source",
    "new_treatment_recommendation",
    "medication_or_dose_change",
    "patient_message_without_approval",
    "sensitive_content_not_required_for_workflow",
    "unsupported_clinical_claim"
  ],
  "default_action": "flag_for_review"
}

If an output attempts to add clinical content that was not present in the source note, Myukura can flag it, block it, or require additional professional review depending on the workflow configuration.

Operational Connectors

Roadmap

Myukura can connect clinical documentation with operational workflows such as email, calendar, forms, task management and patient communication channels. These integrations are designed to support clinical operations, not replace clinical judgment.

{
  "gmail": "draft_only_until_approved",
  "calendar": "pending_approval",
  "whatsapp_business": "preview_until_approved",
  "forms": "reviewable_before_sending",
  "billing_or_admin_tools": "task_based",
  "external_actions": "approval_required"
}

Examples:

  • Prepare an email draft after a consultation.
  • Create a follow-up task for reception.
  • Prepare a WhatsApp message preview.
  • Suggest a calendar reminder.
  • Generate a form request.
  • Track pending administrative actions.

No external action is designed to be executed automatically without professional or authorized user approval.

More Than One Control Layer

The examples above are simplified. Myukura's governance approach is not based on a single prompt or a single setting. It combines product boundaries, technical controls, safety gates, audit logs, professional approval and privacy-aware processing modes.

{
  "governance_layers": [
    "product_boundaries",
    "privacy_controls",
    "role_based_access",
    "human_review",
    "safety_gates",
    "audit_logs",
    "data_minimisation",
    "raw_audio_non_retention",
    "connector_permissions",
    "deployment_mode_controls"
  ]
}

Myukura's goal is to make clinical work faster, clearer and better organized while keeping healthcare professionals in control.

© 2026 Myukura · MYULAB INNOVATION© 2025 NVIDIA, the NVIDIA logo, and NVIDIA Inception are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
Privacy PolicyLegal NoticeCookies