SALIENCEAuditable intelligence

Auditable intelligence

Find where to look first.

SALIENCE turns complex data and documents into traceable signals to support human decisions.

AI organizes · Rules compute · Experts decide

The problem

There is no shortage of data. Clarity is scarce.

Spreadsheets, portals, documents and systems often describe the same process in different ways. SALIENCE connects those sources, identifies relevant patterns and shows the origin of every signal.

How it works

Three steps, one auditable trail.

01

Connect sources

Spreadsheets, APIs, PDFs, portals and documents.

02

Identify signals

Deterministic rules, semantic relations, variations and divergences.

03

Produce evidence

Every occurrence carries source, rule, intensity, confidence, possible explanations and review status.

The central unit

The Evidence Card.

No signal exists without an origin. Every occurrence becomes a card with source, applied rule, intensity, confidence, alternative explanations and the state of human review — audit-ready.

A signal is not an accusation. It is an analysis priority.

SAL-FOL-2026-0148awaiting validation

Atypical monthly variation

Intensity
Confidencehigh
Related sources3
Applied ruleFOL-VAR-07 · v1.2
Why was it flagged?

Variation above the historical behavior of the comparable group.

What still needs confirmation?

Retroactive event, functional change or exceptional payment.

Next action

Review by the payroll specialist in charge.

Signal ≠ accusation. This card organizes evidence for human review — it does not conclude, accuse or determine wrongdoing.
collection: illustrative examplereview: pending

Solutions

Applications of the engine.

in development

Payroll

Monthly data, pay items and variations turned into verifiable signals.

in development

Procurement & Contracts

Suppliers, scopes, prices, deadlines and documents analyzed together.

in development

Data Reconciliation

Comparison across ERP, publications, reports and files sent to oversight bodies.

public demonstration

Document Intelligence

Triage, classification, linking and prioritization of large archives.

The readiness labels are real: today, the engine's public demonstration is the PURSUE case in the SALIENCE Lab. The other modules are in development.

One engine, different domains

The subject changes. The principle stays.

Organize sources, measure signals, document the evidence and preserve the human decision.

in development

Payroll

Monthly data, pay items and variations turned into traceable signals.

public demonstration

PURSUE/UAP

331 file records, 7 signals, recorded provenance and an adjustable ranking — over the official war.gov manifest.

See the case →

SALIENCE Lab · public proof

The method, open for inspection.

Numbers computed at load time from the published dataset — nothing typed by hand.

331

file records (334 in the official manifest)

R01–04

public releases · 7 signals per record

89.3%

within-1 agreement in the blind calibration (§7)

PT/EN

shareable configuration · traceable official sources

Services

Ways to work together.

SALIENCE SCAN

Fixed-scope diagnostic.

SALIENCE MONITOR

Recurring monitoring of new occurrences and changes.

SALIENCE CONCILIA

Comparison across databases, documents and publications.

SALIENCE PARTNERS

Integration, methodology, implementation and joint operation.

Methodology & limits

SALIENCE does not accuse, does not convict and does not replace the expert.

It organizes evidence and makes the analysis reproducible. SALIENCE does not determine fraud and does not replace auditors, accountants, controllers, lawyers or other experts — it organizes sources, applies rules, identifies patterns and produces traceable evidence for human review.

  • — Every signal needs an origin.
  • — A lead is not a verdict.
  • — Traceability before conclusion.
  • — AI organizes. The expert interprets.

About

Anderson Tortela

Anderson Tortela combines experience in operations, contracts, procurement, corporate support and systems building. He created SALIENCE to turn scattered data into clearer, traceable and accountable decisions.

Is there something in your data no one can see clearly?

Contact channel being set up. Meanwhile, the method is documented in detail — and the public case is open for inspection.