We are currently building something exciting and will be launching soon. Registration is temporarily closed. Launching soon — registration temporarily closed.

Independent assurance

Scoped reviews, dated evidence.

Three separate review scopes - because a web penetration test cannot prove model quality, cryptographic correctness, or non-retention. Each entry shows what was examined, by whom, when, and what it does not prove. We publish scope summaries and limitations, not a “secure” badge.

Review scopes

Application & platform security

Scoped - not yet performed

Conventional offensive-security review of the hosted service — but scoped past the generic web checklist.

This scope is defined and budgeted but no engagement has completed yet. Related claims below are labeled self-assessed until then.

Evidence integrity & privacy properties

Scoped - not yet performed

Cryptographic and data-lifecycle review — the properties a pen test cannot see.

This scope is defined and budgeted but no engagement has completed yet. Related claims below are labeled self-assessed until then.

Detection behavior & model quality

Scoped - not yet performed

Behavioral review of the analyzer itself — held-out quality, not a web scan.

This scope is defined and budgeted but no engagement has completed yet. Related claims below are labeled self-assessed until then.

Reviewer access is granted per scope - staged environments, and source/configuration review under NDA where the assurance question requires it. Production access happens only under explicit written authorization. Report contents and finding details are confidential; what we publish is the dated scope, the assessor, and the limitations.

What our assurance claims map to

Every advertised assurance claim names the scope that examines it, the review date, and the remaining limitations. “Self-assessed” means internal review only - we do not imply independent coverage we do not have.

Self-assessed - independent review pending Evidence integrity & privacy properties scope

Signed receipts are verifiable offline and bind the response, configuration, and input commitment

Advertised on: /verify, /docs/verification

Self-assessed - independent review pending Evidence integrity & privacy properties scope

Submitted analysis text is not persisted after the response completes

Advertised on: /zero-retention, /trust-center

Self-assessed - independent review pending Evidence integrity & privacy properties scope

Account deletion is durable — restores cannot resurrect deleted data

Advertised on: /legal/data-rights, /trust-center

Self-assessed - independent review pending Application & platform security scope

Customer accounts are isolated — no cross-tenant data or entitlement leakage

Advertised on: /security, /trust-center

Self-assessed - independent review pending Application & platform security scope

Service degrades fairly under overload rather than failing unpredictably

Advertised on: /docs/api (error semantics), /status

Self-assessed - independent review pending Detection behavior & model quality scope

Detection quality is measured on held-out suites under adverse conditions, including honest abstention

Advertised on: /benchmarks, /docs/evaluation

Self-assessed - independent review pending Detection behavior & model quality scope

A pinned configuration reproduces results; version changes are explicit, never silent

Advertised on: /verify, /docs/configuration-pinning

Security questions, or under NDA and need reviewer-level detail? Report a security issue or contact us.