01 / CLAIM
The question examined
Claim, requirements, control context, system state and criteria, with the intended decision and audience.
BOUNDED ASSURANCE ENGAGEMENTS
What conclusion can the evidence responsibly support?
Examine a defined claim about an AI system through a traceable chain of requirements, controls, evidence, provenance, verification and findings.
The report states the conclusion the examined record supports, the unresolved questions and the boundary within which others may rely on it.
Explore the engagement scope →Who it serves
For teams making decisions that depend on an AI system claim.
Security, AI governance, risk and leadership teams seeking a traceable basis for relying on a specific claim in a stated operating context.
Agreed before work
Define the examination boundary.
| Scope element | What we agree |
|---|---|
| Subject | System, version, configuration, environment and accountable owner. |
| Question | Claim, control or decision to be examined; applicable criteria. |
| Evidence | Available records, permissions, provenance, exclusions and observation period. |
| Delivery | Named responsibilities, duration, client effort, fees, report recipients and permitted reliance. |
Deliverable
What you receive.
An agreed report stating the subject, claim, criteria, examination basis, findings, exclusions and explicit reliance boundaries.
02 / BASIS
The chain behind the finding
Evidence references, provenance, verification performed, conflicting records and material gaps.
03 / CONCLUSION
The permitted reliance
What the record supports, what remains unresolved, exclusions and the limits on using the report.
A commissioned engagement does not guarantee a favourable conclusion. If the evidence cannot support the claim, the report states that limitation.
After the report
Proceed, refine or stop.
The report informs your next decision. Any further examination or remediation support requires a separately agreed scope; there is no obligation to continue.
A changed system, configuration or operating period may require renewed examination. The report does not automatically extend to future behaviour.
Notably Absent
Keep the conclusion inside the evidence.
No absolute safety or security guarantee, regulatory endorsement, certification or assurance about future or unexamined behaviour.
- No claim that a mapping, published framework or source tier establishes the conclusion.
- No missing record treated as evidence that an event did not occur.
- No accreditation, statutory audit status or external acceptance inferred from the engagement name.
Common questions
Before you enquire.
How does this differ from the other assessment options?
Evidence Readiness examines whether records are suitable for a defined examination. Bounded Control Assessment examines selected controls against stated criteria. This engagement connects the examination to a specified claim and an explicit conclusion and reliance boundary. The appropriate route depends on the question and evidence, rather than an assumed progression.
How much does it cost and how long does it take?
Fees, duration and client effort are quoted after commercial qualification. They depend on scope, access and the condition of the records.
Who performs the work?
The proposal identifies responsibility and the competence needed. Relevant conflicts are screened before acceptance; unmanageable conflicts exclude the work.
Can we send records now?
No confidential records or credentials should accompany an initial enquiry. Evidence exchange follows agreed permissions and secure handling arrangements.
Initial discussions establish fit and commercial scope. Technical analysis, study design, validation and advisory work begin only under an agreed paid scope, after independence and data-handling checks and payment. Please do not send confidential records, personal datasets or credentials with your first enquiry.
Start with the claim, the intended decision and a non-confidential description of the records available.