Enterprise

Run agnt8x behind your own firewall.

Deploy agnt8x inside your own AWS, Azure or Google Cloud account, from a signed image your team deploys. Your data, your keys and your models stay in your perimeter, and EightX has no standing access.

Discuss a pilot Deployment options

Deployment

Three ways to run it, one set of controls

The same identity, policy and audit controls apply wherever agnt8x runs, so you can start small and move inside your perimeter without rebuilding anything.

SaaS

The quickest start. agnt8x hosts the platform, in-region, with your organisation isolated from every other client.

Tenant workspace

A dedicated workspace for your organisation, with your own identity provider, keys and data residency.

EMBASSY, inside your cloud

The whole platform in your own cloud account behind your firewall. No egress you have not approved, customer-managed keys, and no standing access for EightX.

Technology

What runs inside your perimeter

Each layer does one job, and each can be inspected. Models are certified per task on your own test cases. If a certificate lapses, the loop stops rather than guessing.

Request assurance documentation

Your agents and loops

Prebuilt agentsWorkflow loopsYour own agents

Control plane

Passport identitySENTINEL policyVERITAS audit

Models and tools

Certified modelsCertified failoverMCP broker

Your infrastructure

Your cloud accountYour keysYour identity provider

Security and assurance

Built for the questions your risk team will ask

Who is this agent?

Every agent carries a Passport: a verified identity and a named owner, checked before it acts or spends.

What is it allowed to do?

SENTINEL checks policy before and during every action. Connections to your systems are read-only unless a write is approved.

What did it do, and why?

VERITAS keeps an immutable audit trail of every step, with the inputs, the model used, the rule applied and the person who approved it. Any run can be replayed.

Which model made the call?

Models are certified per task on your own test cases, run in-region, and fail over only to another model certified for the same task.

Regulation and standards

Aligned to ISO/IEC 42001

agnt8x is focused on helping clients meet ISO/IEC 42001, the international standard for AI management systems, and the AI rules that apply to them across Australia, the United Kingdom, the United States and the European Union.

EU AI Act and the US NIST AI Risk Management Framework
APRA CPS 230 and CPS 234 in Australia
FCA Consumer Duty and SM&CR in the United Kingdom
Financial regulators' expectations on model and operational risk

How it starts

A four-week pilot on one of your processes

Choose a process

Pick one regulated process. We start from your procedure, your rules and your test cases.

Build and certify

We install a prebuilt workflow loop or our agents build one from your procedure, then certify it on your cases.

Run it with your people

Your team runs the loop, approves where the risk calls for it, and reviews the audit trail.

Decide

You keep the loop either way. It is described in the open agent manifest and belongs to you.

Access to agnt8x is by invitation. Pilots run at no cost to the client.

Frequently asked

What enterprise teams ask us

Can agnt8x run inside our own cloud?

Yes. agnt8x runs as SaaS, in a tenant workspace, or inside your own AWS, Azure or Google Cloud account behind your firewall (EMBASSY). Inside your cloud it is deployed from a signed image your team deploys, with customer-managed keys and no standing access for EightX.

See deployment options →

How does agnt8x keep a record of what agents do?

VERITAS keeps an immutable audit trail of every step an agent or workflow loop takes: the inputs, the rule or model used, the result and the person who approved it. Any run can be replayed step by step.

Security and assurance →

Which AI models does agnt8x use?

agnt8x is not tied to any model provider. Models are certified per task on your own test cases and run in-region. Failover goes only to another model certified for the same task, and a loop stops rather than guessing if a certificate lapses.

Which standards and regulations does agnt8x support?

agnt8x is focused on helping clients meet ISO/IEC 42001, the international standard for AI management systems, and the AI rules that apply to them, including the EU AI Act, the NIST AI Risk Management Framework, APRA CPS 230 and CPS 234, and the FCA Consumer Duty and SM&CR.

How does a pilot work?

Access starts with a four-week pilot on one of your processes, at no cost. We install a prebuilt workflow loop or build one from your procedure, certify it on your cases, and your team runs it with approvals and a full audit trail.

Discuss a pilot →

Who can sign up for agnt8x?

Access is by invitation. Organisations start with a conversation with our team; people in a client organisation are invited by that organisation’s administrator.

Ready to run agents inside your firewall?

Talk to our team about architecture, assurance documentation and a pilot in your environment.