Every consequential figure should survive scrutiny.
Model-risk frameworks were built for models. AI agents now select the inputs to calculations and carry the results into decisions, reports and client communications.
Certanum is being built so that every consequential figure is computed from authoritative source data, with evidence a validator, auditor or examiner can re-run.
A figure can be correct, the tool call can be logged and the action can be permitted — while the number rests on an input the model supplied.
An AI agent decides whether to run a calculation, chooses which balances, rates, prices or periods go into it, and reports the result. Three of those four steps rest on model judgement. A wrong share class, a stale rate or a skipped calculation still produces a figure that looks right.
Output review and aggregate accuracy testing cannot establish how a particular released figure was produced. Execution controls establish whether an action was permitted, not what supports the number it produced.
Where consequential figures arise.
A figure is consequential when it is acted upon, recorded, submitted or defended.
Existing frameworks already expect consequential figures to be evidenced.
Descriptions of regulatory and policy instruments are indicative, are not legal advice, and should be confirmed with qualified counsel. Certanum confers compliance with nothing. It produces evidence an existing assurance, validation or audit process may require.
For every consequential figure: computed, traceable, reproducible — or withheld.
Computed. The figure is produced by declared, deterministic software — not generated by the model.
Traceable. Every operand entering the computation resolves to an authoritative source value, or to a declared, immutable artifact bound by digest.
Reproducible. A third party can re-execute the declared computation over the recorded operands and obtain the released figure.
Withheld when unsupported. An operand that cannot be resolved does not degrade the answer. It withholds it.
The architecture is model-agnostic and designed to run inside the institution’s own environment. It produces evidence for existing model-risk, validation and audit processes; it does not replace them.
One workflow. One class of consequential figure. Success criteria agreed before we begin. Certanum funds the initial pilot work.