← The arsenal
ABH-09Use case

Lekha

लेखा

the ledger · the reckoningFinancial-Crime & Money-Laundering Detection

Banking & financeLaw enforcementRegulators

What if your investigators could skip ten thousand false alerts and open the fifty cases that are actually laundering money?

An anti-laundering team faces a flood of alerts, almost all false, and the real laundering hides in the noise — and the data is too sensitive to send anywhere.

The situation

A financial-intelligence unit

Lekha · scenario playing
AML CONSOLE · 4.1M TXNS / DAY10,000+ raw alertsMULE NETWORKscore 0.94 · court-readySTRUCTURINGscore 0.89 · court-readyRAPID LAYERINGscore 0.81 · court-readyON THE UNIT'S OWN SERVERS · DATA NEVER LEAVES

Millions of transactions a day, and a handful are layering dirty money through mule accounts. Lekha learns the shape of normal on the unit's own servers, then surfaces the few patterns that don't fit — structuring here, a sudden mule network there — each one ranked and backed by the trail, ready to act on.

How it plays out

Step by step

  1. 01

    It learns 'normal'

    Lekha builds a picture of ordinary transaction behaviour from the institution's own data — nothing sent out.

  2. 02

    It spots the anomaly

    Structuring, rapid layering and mule-account fan-out stand out against that baseline.

  3. 03

    It ranks the leads

    Instead of ten thousand alerts, investigators get a short, ranked list of the strongest cases.

  4. 04

    It shows the trail

    Each flag comes with the evidence behind it — explainable enough to put before a court.

ABH-09 · readout live
TRANSACTION GRAPHSOURCEEXIT⚠ MULE NETWORK · STRUCTURING

The system itself

Under the bonnet

Reads transaction and account data, learns what normal looks like, then flags the structuring, layering and mule-network patterns that signal laundering — ranking leads by strength and showing the trail behind each, so investigators chase cases, not noise.

FINANCIAL-CRIMEAUDITABLEON-PREM

What it means for you

The bottom line

  • Cuts false alerts to a ranked shortlist
  • Surfaces structuring, layering & mule networks
  • Every flag is court-ready evidence
  • Sensitive financial data never leaves the institution