Every material conclusion in a Briavix diagnostic should connect back to a source document, a validated field and a documented model assumption.
Invoices, contracts, AR aging, shipment and FX data as provided.
Structured data pulled from those documents, field by field.
Critical fields confirmed by a human analyst before use.
Deterministic calculations run only on validated inputs.
Structured fields are extracted from source documents using AI-assisted tooling. Extraction proposes values — it does not finalise them. Every critical field is reviewed by a human analyst before it reaches the model: this is a tool in the workflow, not the product’s value on its own.
An analyst reviews extracted fields against source documents, resolves discrepancies, and flags anything that cannot be confirmed. Only validated fields feed the model.
Margin-at-Risk and scenario figures are produced by a fixed, versioned calculation — not a generative or probabilistic output. The same validated inputs always produce the same result.
Each report states a confidence level reflecting data completeness and validation status — never hidden behind the headline number.
Every report states the model version it was generated with.
The diagnostic proceeds with available data. Missing fields are flagged explicitly in the report, and the confidence level reflects the gap rather than masking it.
No — calculations are deterministic. The same validated inputs produce the same output, and every report states its model version.
A human analyst reviews critical extracted fields before they reach the model — see Analyst validation below.
Yes — every material figure traces back to a source document and a validated field, per the traceability model below.
Discuss your documents and what a diagnostic would show.