Auraium

Enterprise-grade audit intelligence. Finally affordable for every CA firm.

00:00:00:00

launches on 1st April 2026

Join Waitlist See how it works

Why Auraium?

Rule-Based

Fast, deterministic checks for known suspicious patterns (duplicates, blacklisted accounts, threshold breaches).

Statistical

Robust outlier scoring using distribution-aware math for volume, frequency and velocity deviations.

ML Inference

Models identify complex patterns and subtle fraud that rules miss — continuously improving with data.

5-Stage Detection Pipeline

A clean, customer-friendly summary of how Auraium analyzes transactions. Every stage emits a score + human explanation. High-risk transactions advance through all stages and are re-evaluated by our LLM-based fraud validator.

  1. Rule-Based Check
    Fast deterministic checks (new beneficiary, blacklist match, duplicate detection, threshold breach). Lowest latency. Perfect for eliminating obvious fraud.
  2. Statistical Check
    Outlier scoring using z-scores, IQR, rolling percentiles, velocity patterns, amount deviations, and frequency shifts.
  3. ML Check (Inference)
    ML model computes fraud probability, extracts contributing features, and generates a short model-based justification.
  4. Decision Engine
    Consumes CSV outputs from rule, statistical, and ML checks. Fields used:
    • rule_score
    • rule_reason
    • stat_score
    • stat_reason
    • ml_score
    • ml_reason
    • combined_score
    Applies ensemble fusion to determine final risk level and forward high-risk transactions to the LLM validator.
  5. LLM Fraud Recheck
    A privacy-restricted LLM receives pseudonymized transaction context. It gives a natural-language, human-audit-ready final verdict and produces a short, clear explanation of why the transaction may be fraudulent.

Full Technical Pipeline (8-Stage)

The transparent, auditor-ready breakdown of how Auraium processes each transaction.

Show complete 8-stage flow
  1. Ingest — Load & standardize CSV/Excel.
  2. Validate — Schema validation & column mapping (LLM optional for column names only).
  3. Rule Check — Deterministic fraud rules.
  4. Statistical Check — Outlier detection per metric.
  5. ML Check — Probability + feature contributions.
  6. Decision Engine — Rule/stat/ML fusion → combined score → ranked suspicious transactions.
  7. Fraud Validator — Post-decision business rules & external validations.
  8. LLM Explainer — Final natural-language verdict + explanation.

Pipeline timeline

Visual flow that shows how transactions travel through the system.

Rule Stat ML Decision LLM

Live pipeline flow

Scroll horizontally through the pipeline stages with subtle animation.

Rule-Based Check

Deterministic detectors that catch known patterns quickly.

Statistical Check

Outlier models and rolling-window baselines.

ML Inference

Model score, top features, shap summaries.

Decision Engine

CSV fusion, combined score, ranked candidates.

LLM Recheck

Pseudonymized final verdict and natural language explanation.

Plain English explanations (example)

Transaction ID: trxn_00012345
Amount: ₹1,20,000
Date: 2025-11-26
Final verdict: FLAGGED

Privacy by design

We minimize data shared with external services and always prefer on-premise or private inference for sensitive fields.

Minimal fields to API

Only necessary, non-identifying fields are sent to third-party services or LLMs for mapping: column headers, transaction metadata tags, and hashed IDs. Raw PII (names, account numbers) is never sent without being obfuscated or hashed — and only when explicitly allowed.

Local-first inference

Models run in your controlled environment when possible. If cloud inference is used, data is pseudonymized and logged with strict retention policies.

Auditable trace

Every API call is logged with a minimal bleed of fields and includes a reason code. This provides traceability while protecting privacy.

FAQ

What happens while mapping fails?

If automatic mapping fails, Auraium will (optionally) call an internal LLM helper with only column names — not values — to suggest mappings. You always see suggested mappings before they’re applied.

How do we measure precision / false positives?

Auraium tracks precision, recall, and false positive rate per rule and per model. You can view suite-level metrics on the dashboard and set thresholds to meet your tolerance.

Can I opt-out of cloud ML?

Yes — Auraium supports on-premise models and private inference. Cloud ML is optional and used only after explicit configuration.

Ready to try clearer, safer detection?

Get started Contact sales