Foundation · Roadmap

Where approved models run, on your terms.

A planned local runtime for approved models, with jurisdiction, model routes and audit evidence defined for each deployment.

The evaluator's problem

If the model runs elsewhere, sovereignty is a claim, not a fact.

Where inference happens affects whether data is truly under your control. The runtime is intended to keep execution within the boundary defined for the deployment.

01

Where does inference happen?

On an organisation-approved local runtime, inside your boundary.

02

Which models can run?

Only models your organisation has approved, a signed, known set.

03

Can we audit it?

Execution is governed and recorded, not opaque.

Capabilities · Accessible by design

Execution you control.

The runtime is where the sovereignty promise is kept or lost.

Local execution

Approved models can run inside a customer defined boundary.

Signed model policy

Only an approved, signed set of models can run.

Approved routes

Reasoning follows the model routes approved for the deployment, with access and jurisdiction recorded.

Governed & recorded

Execution is auditable.

Where it fits

Where the Runtime fits.

The engine beneath intelligence, standing on the data foundation.

Governance & compliance

Sovereignty, where it is decided.

Inference can run on an approved local runtime, with the model set, jurisdiction and evidence recorded for the deployment.

Customer controlled data boundary Human approval where it matters Reviewable activity record Designed to support AI Act governance Designed to support GDPR obligations

Next step

Run one workload locally.

We stand up one workload and document where inference runs, which routes are approved and how the boundary is enforced.