EVIDENCE & TRUST

Precise claims are part of responsible AI.

SundAI separates verified facts, source-backed guidance, professional interpretation, illustrative examples and commercial availability so readers can see what is documented, what is guidance and what remains context-dependent.

CLAIM LABELS

What each type of statement means.

Verified fact

Documented public information

Company registration, publication dates, official regulatory timelines and public technical documentation are linked to the underlying source whenever practical.

Source-backed guidance

Interpretation with citations

Guides explain how official requirements and frameworks can translate into operational decisions. They are not presented as regulator, standards-body or legal advice.

Illustrative

Delivery scenarios

Examples show how an engagement could be structured. They are explicitly not named client references, completed engagements or guaranteed outcomes.

Measured outcome

Published only with evidence

Performance, savings, adoption, security or risk-reduction claims should be published only when there is a defined baseline, measurement method and permission to disclose the result.

Relationship

Partners are not inferred

Using a technology platform, holding a certification, attending a programme or linking to another organisation does not by itself mean a formal partnership.

Certification

Scope matters

SundAI does not describe the organisation as ISO-certified unless a current certificate with a verifiable scope is published. Individual education and organisational certification are kept distinct.

Availability

Published options are not current engagements

A service page or indicative price describes a possible organisation-level commercial format. It does not by itself establish a current client engagement, immediate availability, reserved capacity, a fixed staffing model, personal delivery by a named individual, hours worked, client volume or revenue.

EVIDENCE HIERARCHY

Primary sources first, interpretation second.

01

Primary authority

Legislation, regulators, standards bodies, official registries and first-party technical documentation.

02

Direct project evidence

Approved deliverables, test records, measurements, contracts, source code and other first-party artifacts where publication is permitted.

03

Secondary context

Reputable research and analysis can add context, but should not replace a primary source for a legal or standards claim.

04

SundAI interpretation

Practical recommendations are labelled and should be tested against the organisation’s role, system, sector and applicable law.

SERVICE AVAILABILITY

Availability is confirmed only when an engagement is agreed.

Commercial accuracy: Scope, contributors, delivery model, timetable and capacity are confirmed separately for each enquiry before work is accepted. A public service description is not evidence that work has been sold, started, delivered or personally undertaken by any named person.

PUBLIC EVIDENCE

Useful places to verify SundAI and its published work.

Company registry

RIIMON HOLDINGS LTD, company no. 13813344. Registry status can change, so the official register is the source of truth.

Open Companies House

Public repository

Selected website and technical materials are published in the SundAI GitHub repository, providing a public change history for those artifacts.

Open GitHub

Methodology

The delivery model explains how governance, security, human oversight, evidence and lifecycle review fit together.

Read methodology

AI Control Matrix

A source-linked practical mapping aid across the EU AI Act, ISO/IEC 42001, ISO/IEC 27001 and NIST AI RMF.

Open matrix

Readiness Brief 2026

A dated evidence brief based on official EU, ISO and NIST sources, with a transparent self-assessment rubric.

Read brief

Illustrative case studies

Examples are labelled as illustrative until named client work and measured outcomes can be published with permission and evidence.

View scenarios
Review policy: Date-sensitive pages are reviewed when material information changes. Historical wording is not silently converted into a stronger claim. Corrections preserve a clear distinction between what was previously published and what is currently supported.

WHAT SUNDAI DOES NOT CLAIM

Trust is also defined by what is left unsaid.

No invented clients or testimonials

Client names, logos, quotes and project results are not published without a factual basis and appropriate permission.

No automatic compliance guarantee

Templates, checklists and framework mappings can support compliance work but do not prove that a specific AI system or organisation complies with law.

No implied regulator or vendor endorsement

Citing official sources or working with a platform does not mean the regulator, standards body or vendor endorses SundAI.

No inference of individual working time or availability

An organisational role, publication, service page, price or enquiry mechanism does not by itself state a person's working time, personal availability or amount of work performed.