AI solutions, governed end-to-end
Privexa is the protection primitive for clinical AI workflows. Privexa Trace is the governed evidence platform built on top of it. Together they form a complete, reviewed AI operating model.
Who this is for
Clinical Sponsors
Biotech and pharma teams running evidence reviews
CROs
Contract research organisations managing trial workflows
Academic Groups
Research teams needing governed synthetic control work
Diagnostic Vendors
Companies integrating AI into diagnostic pipelines
The case for clinical AI
From diagnostic imaging analysis to administrative automation and clinical decision support, AI is delivering measurable improvements in patient outcomes and operational efficiency. New Zealand healthcare organisations are at a pivotal moment — early adopters are gaining significant advantages, while those waiting face increasing pressure from staff, leadership, and patients who expect modern tools.
Translyx helps organisations identify, evaluate, and implement AI solutions that are clinically validated, fit for purpose, and safe to deploy in New Zealand's healthcare environment. That responsibility extends beyond privacy alone.
For healthcare and research teams, responsible AI also requires a governed workflow around evidence assembly, review gates, and traceable outputs. That is where Privexa and Privexa Trace come in.
Privexa
Protected AI privacy layer
Protects patient and clinical data before it reaches AI systems or cloud workflows.
Privexa Trace
Reviewer-gated evidence workflow
Structures synthetic control and evidence operations with manual benchmark comparison, review gates, and package-ready outputs.
The data privacy challenge
Most AI platforms require data to leave your environment. Patient names, NHI numbers, dates of birth, clinical notes, and diagnostic details are routinely transmitted to external systems when staff use AI tools — often without explicit awareness of the privacy implications.
Under the New Zealand Privacy Act 2020, health information is among the most sensitive categories of personal data. Organisations have a legal and ethical obligation to ensure that AI adoption does not create exposure that existing governance frameworks were not designed to handle.
The question is not whether to adopt AI. The question is how to adopt it without creating privacy risk that undermines patient trust and organisational compliance.
Privexa — Enterprise AI privacy platform
Live at app.privexa.co
Privexa is a live enterprise AI privacy platform — available now at app.privexa.co. It sits between your data and the AI systems your teams depend on, with five integrated protection layers covering LLM workflows, cloud pipelines, clinical documentation, and compliance.
Before any data reaches an AI system, Privexa detects and replaces sensitive fields — patient names, NHI numbers, dates of birth, clinical identifiers, and custom entity types — with safe tokens. AI systems operate on the tokens. Original data never leaves your environment. When AI responses return, Privexa restores original context seamlessly and invisibly.
Secure LLM Gateway
PII is intercepted and replaced with tokens before any message reaches an LLM. Responses are automatically restored with original context. AI workflows operate normally — patient data never leaves your environment.
Real-time PII detection, token replacement, automatic response restoration.
Cloud Shield
Field-level protection for cloud data pipelines. Sensitive fields are tokenised before reaching AWS, Azure, GCP, Snowflake, or Oracle. The local mapping vault is reversible only inside your perimeter — inaccessible to any external system.
Raw PII never reaches your cloud platforms.
Privexa Scribe
Clinical consultation recording with full PHI protection. Record the full session and transcribe on stop, or enable live relay for in-session visibility. Privexa de-identifies PHI, runs the note pipeline, restores identities locally, and returns a clinician-ready draft with review controls.
SOAP notes, referrals, discharge summaries, ICD support — PHI stays inside your boundary.
Document Intelligence
Analyse contracts, clinical records, and reports with AI — without exposing a single name, number, or identifier to any external model. Structured summaries, compliance scans, risk assessments, and entity mapping, all privacy-first.
HIPAA, SOC 2, GDPR compliance scanning and risk assessment on sensitive documents.
Governance & Compliance
HIPAA and GDPR compliance monitoring, benchmark testing, data lifecycle controls, and immutable audit trails — built into the platform architecture, not added as optional modules.
Audit trails, access controls, and data handling designed for SOC 2 alignment.
Privexa Trace — governed evidence workflows for synthetic control work
Protecting patient data is the starting point. What follows — how evidence is assembled, reviewed, and packaged — determines whether clinical AI is actually governable in a regulatory context.
Privexa Trace addresses that second layer: reviewer-gated synthetic control arm workflows for biotech, pharma, CRO, and clinical trial office teams. Reviewer sign-off is required before SCA generation. Limitations are visible in every output. Packages are structured for submission from the outset.
Reviewer sign-off before SCA generation
Trace keeps a hard review gate in the workflow so statistical generation does not outrun governance.
Manual and synthetic comparison in one workspace
Specialist-entered manual treatment and manual control benchmark arms are assessed alongside synthetic workflow outputs.
Traceability, lineage, and submission packaging
Workflow context, visible limitations, and submission-oriented output remain part of the product story — not an afterthought.
Worked example
One question, three paths, one traceable package
A team defines the clinical question → a specialist builds manual treatment and control arms → Trace structures comparable evidence → reviewer signs off before SCA generation → the final package carries lineage, rationale, and visible limitations forward.
Designed for regulated healthcare
Privexa is built with New Zealand's regulatory environment in mind:
HIPAA-aligned
Patient data never reaches external AI systems. PHI stays inside your boundary.
GDPR-ready
Personal data minimisation enforced at the API layer. Consent-aware processing.
NZ Privacy Act 2020
Designed to meet NZ information privacy principles for AI-assisted workflows.
SOC 2 oriented
Audit trails, access controls, and data handling designed for SOC 2 alignment.
Common questions
What is the difference between Privexa and Privexa Trace?
Privexa is the data-protection layer — it ensures sensitive clinical data never reaches AI systems in identifiable form. Privexa Trace is the evidence workflow product — it governs reviewer-gated synthetic control arm workflows with lineage and submission packaging. They solve different problems in the same AI governance stack.
Does Privexa Trace imply regulatory approval?
No. Privexa Trace is designed to support submission-oriented packaging and reviewed governance workflows. It does not claim or imply regulatory endorsement from FDA, EMA, Medsafe, or any other body. Translyx is transparent about this on every product page.
Who is Privexa Trace designed for?
Biotech, pharma, CRO, and clinical-trial office teams who need governed synthetic control arm workflows, visible review gates, manual benchmark comparison, and traceable output that can be defended in a submission context.
Deploy clinical AI with the governance your organisation requires
Translyx can walk your team through Privexa's data protection architecture and Privexa Trace's reviewer-gated evidence workflow — and identify where each fits your clinical AI roadmap.