Enterprise AI assurancePrivate preview available by request

    Know what AI systems you run. Prove how they are controlled.

    NVX connects AI systems and agents to their purpose, owners, risks, controls, evidence and approval state—giving enterprises a defensible view of how AI is governed.

    Assurance map · illustrative example

    Approval pending

    AI application

    Customer Resolution Agent

    Controls: 6 linked · 5 operating

    Needs attention

    Each risk maps to controls. A control without evidence is shown as not yet proven.

    The control gap

    AI moves into operations faster than oversight can follow.

    01

    AI is hard to see

    Teams adopt models, copilots and agents faster than anyone can record them, or say who owns them.

    02

    Risk decisions are scattered

    Risks sit in a register, controls in a policy, test results in an engineering tool and approvals in email.

    03

    Evidence goes stale

    A review proves a system on the day it was signed. Models, prompts and data change the next week.

    A policy is an intention.
    Assurance requires evidence.

    Illustrative product experience

    One AI system, from first record to approval.

    Follow one customer resolution agent through four stages. Names and figures are examples, not customer data.

    1. The private preview records and governs assurance results. Automated agent behaviour testing is planned.

    Stage 1 of 4

    Mapped · 3 risks · 6 controls

    Map

    1. Registration

      System
      Customer Resolution Agent
      Stage
      Pilot
    2. Purpose and ownership

      Purpose
      Resolve delivery and refund requests
      Owner
      Head of Customer Operations
    3. Dependency mapping

      Model
      Hosted LLM · pinned version
      Tools
      Refund API, CRM write-back
    4. Risk and control assignment

      R-01
      Refund above policy limit → human approval above A$200
      R-02
      Personal data in prompts → redaction

    Why assurance is different

    Governance tells you what should happen. Assurance shows that it did.

    NVX keeps one connected record across four layers, so a reviewer can trace any approval back to the system, the risk, the control and the evidence behind it. When something changes, the record shows what needs to be checked again.

    1. 01

      Governance context

      Purpose, accountable owners, policies and approval decisions.

    2. 02

      Assurance operations

      Risks, controls, test results, findings, remediation and evidence, linked together.

    3. 03

      AI system graph

      Each application, agent and model, with the tools and data it depends on.

    4. 04

      Enterprise environment

      The cloud, data and business systems your AI already runs in.

    What changes

    From scattered documents to one source of proof.

    Today: AI systems tracked in spreadsheets, if at all

    With NVX: One record per AI system with a named owner

    Today: Risks and controls described in documents

    With NVX: Risks and controls linked to the system they apply to

    Today: Evidence collected by hand before an audit

    With NVX: Evidence kept with the control it supports

    Today: Approvals buried in email threads

    With NVX: Approval decisions recorded and versioned

    What is available

    The private preview, stated plainly.

    NVX is accepting selected enterprise teams into a private preview of its AI governance and assurance workflow. The private preview records and governs assurance results. Automated agent behaviour testing is planned, and it does not yet run in your own cloud. We hold no certifications and show no client logos.

    • AI-system registration, purpose and ownershipFoundation capability — validation underway
    • Risk, control and evidence recordsFoundation capability — validation underway
    • Findings, remediation and approval recordsNVX workflow — validation underway
    • Versioned assurance recordsPlanned for private MVP
    • Deployment inside your own cloud accountPlanned
    • Automated agent behaviour testingPlanned

    NVX is validating how these capabilities operate together as one evidence-backed assurance workflow. Availability may vary during the private preview.

    Enterprise design partner programme

    Shape the platform around your real AI systems.

    We are working with a small number of organisations that run AI in production or are about to. Partners get early access and direct influence over what we build.

    You get

    • Early access to the private preview
    • An assurance map of one live AI system
    • Direct input into the roadmap

    We ask for

    • One AI system to work through together
    • Regular feedback sessions
    • A named owner on your side