Deepcode LLCXpectra is the compliance division of Deepcode
InvestorsNewsroom
01
01Posture
Section 01 · Posture
The file is already written before anyone asks for it
Screening against 50,000+ lists, transaction monitoring over the resolved network, and fraud in the same case file — three lines that nest, and one immutable record with the reasoning, the source and the analyst signature.
A single call opens the whole fan. Each family returns its own weight, and that weight travels with the decision into the file.
Sanctions · OFAC, EU, UN, HMT
PEP tiers & relatives
Adverse media · 10M pages/day
Corporate registries · 195 jurisdictions
Ownership & control graph
Vessels, ports & cargo
Debarment & regulatory warnings
50K+
Lists in scope, refreshed in minutes rather than days.
195
Jurisdictions with searchable corporate registries.
<100ms
Screening response at the 95th percentile.
7 yr
Immutable retention of every decision and its reasoning.
Performance and accuracy figures shown on this site are measured in controlled test environments on representative datasets. Real results vary with data quality, list configuration and volume. The only binding commitments are those stated in the service agreement.
03
03The file
Section 03 · The file
While you read this line, the file is already being written
The subject entity matches EU CFSP 2023/1214 on legal name and registration number. Ownership resolves in three hops to a designated ultimate beneficial owner.
Recommended action: escalate to Tier 2 with the ownership chart attached
awaiting analyst signature
Sign & closeEscalate
04
04Deepsi AI
Deepsi AIThe cognitive layer
Deepsi AI does not score names. It walks relationships.
Ownership, control, address, counterparty and asset. The engine walks every edge until it finds the exposure a name alone never shows, recognises the pattern behind it, and writes down why.
98%
Match accuracy
95%
False positives out
Four typologies the engine detects on the network, not on a name. Each has its own shape, and the shape is what makes it a case.
STRUCTURING
1,284 records · 947 resolved entities · 3,106 edges
Smurfing
Many unrelated depositors paying into one account, each amount deliberately parked under the reporting threshold.
What fires it
Six different depositors on one account in nine days, none of them above the threshold on its own. No single deposit is reportable; the set is.
STRUCTURING
Many unrelated depositors paying into one account, each amount deliberately parked under the reporting threshold.
account or companypersondevice or identifierthe detected shapeWhat fires itthe rest of the network, walked and discarded
Performance and accuracy figures shown on this site are measured in controlled test environments on representative datasets. Real results vary with data quality, list configuration and volume. The only binding commitments are those stated in the service agreement.
04.2
04.2Deepsi AI
04.2 · Patterns
One match is a data point. A pattern is a case.
Deepsi recognises the shape of laundering, not only the name on the list. Four of the typologies it watches continuously:
Money laundering04
hop 1designated
Shell layering
Ownership chains of three or more hops that terminate in a designated party.
3+ hops · 3 jurisdictions
threshold
Structuring
Sequences of transfers deliberately parked just under the reporting threshold.
6 transfers · 94% of threshold
returns to origin
Round-tripping
Funds that leave and return to the same beneficial owner through intermediaries.
cycle closed · 4 hops
declaredmarket
Trade mis-invoicing
Declared cargo value diverging from the market price for the same route, in either direction.
declared 41% below market
Fraud03
new deviceexpected
Account takeover
A session from an unseen device changes the beneficiary and empties the balance before the customer notices.
new device · beneficiary changed in 4 min
23 in1 out
Mule network
Many small deposits from unrelated payers converging on one account that empties the same day.
23 inbound · 1 outbound · same day
118 x $1.00$2,400
Card testing
Bursts of minimum-value authorisations against a card range, looking for the ones that clear.
118 auths · 0.9 s apart · $1.00
Four typologies the engine detects on the network, not on a name. Each has its own shape, and the shape is what makes it a case.
04.3
04.3Deepsi AI
04.3 · Deepsi at work
The engine does not stop at the score
Deepsi proposes the thresholds, works the queue, drafts the file, and learns from every decision your team signs.
04.3.a
Calibrates the system
Proposes matching thresholds per list and per jurisdiction, and projects the effect of each change on recall and on alert volume before you apply it.
threshold 0.62 → 0.68 · −31% alerts
04.3.b
Works the alert queue
Groups duplicate hits into one case, ranks by severity and SLA, and prepares each alert with the evidence an analyst needs to decide in one pass.
1,284 alerts → 96 cases
04.3.c
Writes the report
Drafts the risk narrative with its regulatory citations, the ownership chart and the audit trail, ready for the analyst signature and for an examiner.
SAR-ready · 7-year record
04.3.d
Learns from your decisions
Every dismissal with a reason feeds the scoring model, so the pattern your team rejected today weighs less in tomorrow’s queue.
feedback loop · per tenant
Performance and accuracy figures shown on this site are measured in controlled test environments on representative datasets. Real results vary with data quality, list configuration and volume. The only binding commitments are those stated in the service agreement.
05
05Section 05 · How it is bought
Section 05 · How it is bought
Three lines. Each one contains the one before it.
Most confusion about AML software comes from vendors selling parts as if they were products. Here the three lines nest: you pick where you start, and you move up without changing platform, data or audit record.
Screening
01
Lists and adverse media
For· Teams that need to check a name before onboarding it, and prove later that they checked.
Lists and adverse media
Sanctions, PEP, debarment and regulatory warnings
Adverse media and crime lists
Entity resolution and match scoring
Immutable audit record, 7 years
REST API and CSV bulk upload
AML
02
Screening + transaction monitoring
For· Regulated institutions with an AML programme and a regulator that asks how it works.
Adds over the line before
Transaction monitoring on the resolved network
Smurfing, funnel accounts, layering and mule networks
Rules engine with thresholds, calibration and two-signature approval
Alert queue, case management and regulatory reporting
+ Screening
Suite 360
03
AML + fraud
For· Institutions where the same team, or the one next door, also answers for fraud.
Adds over the line before
Account takeover, mule networks and card testing
Shared device, phone and terminal signals across both
One network and one case file for laundering and fraud
A single alert queue instead of two that do not talk
+ AML
The line is what you can do. The plan is how much volume you do it on — Starter, Advanced or Enterprise. They are two separate choices and you make them independently.
Engineered specifically to process enormous volumes of Sanctions, Watchlist, and Adverse Media data with parallel Continuous Transaction Monitoring.
Integration without friction
We provide API-first flexibility to suit your business. Integrate our intelligence into your preferred workflows via real-time API or batch processing.
Real-time screening capability
Support fast onboarding flows and seamless customer experiences with sub-second entity resolution and risk scoring.
SOC 2 & ISO 27001 infrastructure
Built on Google Cloud, with infrastructure audited for SOC 2 Type II and certified ISO/IEC 27001. Your data travels encrypted with TLS 1.3 and is stored encrypted with AES-256.
Massive scalability
Handle millions of annual screenings with consistent performance. Our distributed mesh architecture grows with your volume.
—
—References
Executive narratives
three customers, on the record
"Xpectra didn't just automate our screening; they defined our compliance DNA. The AI process was surprisingly intuitive."
CMCarolina MéndezChief Compliance Officer, Qualita Group
"I've never seen a platform move this fast without compromising on regulatory precision and audit quality."
RCRoberto CastañedaVP Risk & Compliance, SBS Seguros
"The level of sophistication in their entity resolution engine is lightyears ahead of anything on the market."
All plans includeSOC 2 & ISO 27001 infrastructure on Google CloudEU or US data residency7-year audit record14-day trial, no cardSee the full comparison →
—
Bring us your next examination
We screen your own portfolio during the demo, and show you the file that would exist if an examiner asked for it tomorrow.
04.3.a
04.3.b
04.3.c
04.3.d
