Trust should be inspectable—not implied.

See what the public MatuLabs surface does, what customer-controlled deployments are designed to do, and which claims still require independent evidence.

Current posture

Capability labels separate visible public behavior from engineering evidence, future deployment targets and research direction.

Public website

Live public surface

Proof Lab

Synthetic demonstration

Product workspaces

Private development

ADP specification

Public draft

Trust Rail SDK

Offline developer preview

Global Trust Network

Research only · not live

Direct answers to the questions trust buyers ask

Does the public site use customer data?

No customer data is used in the public Proof Lab. Its workflows and evidence are illustrative and synthetic.

Does MatuLabs silently train on customer work?

No silent adoption is the governing design rule. Any future learning or cross-agent adoption requires an explicit, evidence-bound gate and deployment-specific verification.

Can sensitive work remain inside a customer boundary?

Closed-loop and customer-controlled deployment are product design targets. The exact egress, key custody, retention and operator controls must be proven for each deployed configuration before reliance.

Does AI release final legal, audit, financial or actuarial opinions?

No. MatuLabs prepares reviewable material. Final judgment and release authority remain with authorized professionals.

Is MatuLabs externally certified or production-ready?

This public surface makes no such claim. Internal engineering evidence is not a substitute for independent certification, customer validation or a dedicated production-readiness gate.

Trust architecture

These controls are part of the MatuLabs design and validation program. Their presence here is not an external certification.

Data boundary

Explicit ingress, egress, storage and operator boundaries are verified per deployment.

Identity and authority

Actions are designed to bind agent or human identity to scoped authority and review state.

Evidence lineage

Material claims retain source, scope, version and limitation references.

Replay and incident review

Critical routes are designed for reconstruction, discrepancy review and revocation handling.

Human release

AI-prepared material does not silently become final organizational authority.

Bounded learning

Learning candidates are separated from live adoption and require an evidence-bound promotion decision.

Evidence access model

Public claims stay public. Sensitive technical material, customer evidence and independent reports require a controlled access decision and must not be inferred from this page.

The claim boundary

MatuLabs will not present an internal test as a customer outcome, a design target as a live control, or AI/contractor review as independent certification.