R38 Technologies · Building MuleNet AI

AI-Powered Money Mule Network Detection.

Protecting mid-tier banks, e-money institutions, and payment service providers with graph-based intelligence that sees across transaction networks — not just individual accounts.

Scroll
[ 01 — The problem ]

The mule detection gap.

Money mules are the operational backbone of financial fraud in the UK. Current detection systems were not built to see what moves between institutions.

Money mules are the operational backbone of financial fraud in the UK. They move stolen funds through chains of accounts across multiple institutions, and current detection systems cannot see the full picture because they operate within a single institution's data.

Mid-tier financial institutions — challenger banks, electronic money institutions, and payment service providers — are disproportionately affected. Enterprise AML solutions that offer network-level detection cost upwards of £200,000 annually, placing them out of reach for the firms that need them most.

The result is a detection gap. Neither the sending nor the receiving institution sees the complete mule chain. Funds move through multiple accounts in minutes, and by the time suspicious activity is flagged, the money has already left the network.

R38 Technologies exists to close this gap.

[ 02 — The approach ]

See the ring. Not just the account.

MuleNet AI ingests transaction data via API, constructs a live transaction graph, and runs graph neural networks to surface mule topologies — including accounts predicted to be activated next.

app.mulenet.ai / alerts / cluster-2847
Live
Transaction network — Cluster #2847
Last 24 h · GNN v3.2
Normal Mule ring Predicted
2,847 accounts 6,102 edges 3 clusters Updating
F.01

Graph-based network detection

We use Graph Neural Networks to analyse transaction topology, identifying coordinated mule rings as connected network structures rather than flagging individual transactions in isolation.

GNN
F.02

Forward-looking intelligence

Current solutions ask "was this transaction suspicious?" We ask a fundamentally different question: "which accounts are most likely to be the next link in an emerging mule chain?" This shifts detection from reactive investigation to pre-emptive monitoring.

Novel
F.03

Built for mid-tier institutions

Advanced mule detection should not be a privilege reserved for the largest banks. Our platform is designed as an affordable SaaS layer that integrates alongside existing transaction monitoring infrastructure.

Accessible
[ 03 — How it works ]

Four steps. Human-in-the-loop.

MuleNet AI sits alongside existing transaction monitoring — surfacing what individual-account systems cannot see, then handing every decision back to the compliance officer.

  1. 01
    Integrate

    Connects to existing transaction monitoring infrastructure via secure API. No changes to payment processing.

  2. 02
    Analyse

    Graph Neural Networks continuously analyse transaction patterns, modelling the relationships between accounts.

  3. 03
    Alert

    Generates alerts with risk scores, network visualisations, and explainable AI summaries.

  4. 04
    Decide

    The institution's compliance officer reviews and decides. The platform supports human judgement — it never replaces it.

[ 04 — Why it matters ]

The scale of the gap, in numbers.

Money mule activity in the UK has reached an industrial scale — and current detection systems are not keeping pace. Four figures that define the problem.

S.01
225,000+
Mules identified in the UK in 2024.

A population the size of Portsmouth — and that is only the mules investigators were able to catch.

NCA National Strategic Assessment · 2024
S.02
£1B+
In mule-facilitated fraud, every year.

Losses outpace the UK's annual economic-crime enforcement budget by more than two-to-one.

UK Finance Annual Fraud Report · 2024
S.03
38%
Of mule outflow moves through just three firms.

Risk is not evenly distributed. Detection capability at a handful of mid-tier firms would shift the national picture.

RUSI Independent Research · 2025
S.04
<60 min
The window to freeze funds before they are untraceable.

Coordinated mule chains move stolen funds across multiple institutions in minutes. Most alerts from legacy systems land days later.

FCA Multi-Firm Review · December 2024
[ 05 — Recognised need ]

Recognised need. Active response.

The detection gap R38 Technologies addresses is not a private thesis. It is the consensus view of UK regulators, industry bodies, and independent researchers.

The scale and sophistication of mule networks exploited by organised crime groups has reached critical levels.
UK Finance Annual Fraud Report · 2024
Existing detection approaches are insufficient against coordinated mule operations across institutions.
FCA Multi-Firm Review · December 2024
38% of mule outflow payments are concentrated in just three institutions.
RUSI Independent Research · 2025
[ 06 — Boundaries ]

What R38 is not.

Compliance leaders rightly ask hard questions of any new tool. Here is what we are not — clearing the most common objections before they happen.

Waitlist · Contact us

Join the waitlist.

MuleNet AI is in active development. Join the waitlist to be among the first UK financial institutions, investors, and partners to hear when our platform opens for evaluation, or reach out directly if you'd like to start a conversation now.

✓ Thanks. We've received your message and will be in touch.