AI agent governancePrivate preview available by request

    An AI agent that can act needs limits someone has approved.

    Agents call tools, read data and take actions in business systems. NVX records each agent's permitted tools, data sources and actions alongside the risks, controls and approval that apply to it.

    What changes with agents

    The risk moves from what the model says to what the agent does.

    Tool access

    Each connected tool expands what an agent can change. A refund tool and a read-only lookup carry very different risks.

    Data reach

    Agents often combine sources. The record should show which data an agent can read and which it can write.

    Autonomy level

    Whether a person approves each action, samples actions after the fact, or is only alerted on exceptions.

    Change over time

    New tools, prompts and model versions change behaviour. Approvals need to reference the configuration they covered.

    The agent record

    What NVX records for each agent.

    1. 01

      Purpose and owner

      What the agent is for and who answers for it.

    2. 02

      Tools and permissions

      Each tool the agent can call, with the scope of its permission.

    3. 03

      Data sources

      Systems the agent reads from or writes to, and the sensitivity of that data.

    4. 04

      Human oversight

      Where a person must approve, review or can intervene.

    5. 05

      Approval and conditions

      The decision, its conditions and when it must be revisited.

    Questions

    Frequently asked

    Does NVX block agent actions at runtime?
    Not in the private preview. NVX records approved limits and links them to controls and evidence. Runtime enforcement belongs to your agent platform or gateway.
    Which agent frameworks are supported?
    During the private preview, agents are recorded regardless of framework. Direct integrations are being validated with design partners and are not yet available.
    Is automated agent testing included?
    Automated agent behaviour testing is planned and not yet available.

    See the workflow on one of your AI systems.

    NVX is accepting selected enterprise teams into a private preview of its AI governance and assurance workflow.