Product How it works Capabilities Results Governance FAQs Contact
Log in Request a briefing United Global Funding®

A United Global Funding® product

Screen every investor. Before the round closes.

Seretan® is UGF’s in-house trained large-language-model engine. It screens investors entering Series A and Series B — unwinding ownership, testing source-of-funds narratives, and flagging sanctions, PEP and network risk — before a single dirham, euro or pound is accepted.

In-house trained Human-in-the-loop Sovereign infrastructure Arabic · English · Turkish
Seretan® · risk memo Committee
Series B · incoming LP
Northbridge Holdings Ltd BVI vehicle · three-layer stack · UBO unresolved at intake
Escalate
Entity resolutionUBOs unwound across 3 vehicles, 2 jurisdictions
Resolved
Source of fundsDeclared exit timeline does not support ticket size
Inconsistent
Sanctions / PEP2 false hits auto-adjudicated · 0 true matches
Clear
Network analyticsShared director with a 2024 declined subscription
Linked

Recommendation · Condition or decline

Human decides

Illustrative memo. Seretan® recommends; the investment committee decides.

Built inside UGF® · run on UGF infrastructure
Dubai Istanbul Tallinn Est. 2016

Why venture rounds need a financial-crime defence

An equity stake launders reputation as well as money.

Illicit capital looking for legitimacy arrives as a “sophisticated LP”: a freshly formed vehicle, a fund-of-one, a nominee stack. One tainted name on the cap table can compromise a portfolio company’s banking, a later exit, and the fund’s standing with regulators and LPs.

Legitimacy value

Proceeds become “investment returns.” The investor gains a credible wealth narrative that survives a later source-of-funds review.

Opacity of vehicles

Series A and B investors frequently arrive through new SPVs, nominees and layered LP stacks across multiple jurisdictions.

Sanctions & PEP exposure

Global rounds attract capital from every jurisdiction. Sanctions-evasion and PEP-corruption risk enters the cap table silently without structured screening.

Contagion

One compromised investor can put banking relationships, later exits and the fund’s own regulatory standing at risk. Integrity is a first-order control.

What Seretan® is

Practitioner infrastructure. Not a bolted-on checklist.

UGF does not only research AI against financial crime — it operates it. Seretan® is the private-market application of the same capability stack used to detect networked illicit finance: entity resolution, graph analytics, multilingual NLP and a governed LLM — with a named human on every decision that matters.

01

In-house trained LLMs

Models trained and fine-tuned on UGF’s own deal, KYC and disposition data, augmented with curated public sources — registries, sanctions and PEP lists, adverse media and court records — across the languages of UGF’s markets.

Arabic English Turkish + others
02

Sovereign by construction

Seretan® runs on UGF-controlled infrastructure. Investor data never traverses third-party public LLM services. Logs, model versions and evidence packs stay inside the firm’s perimeter.

03

Human-in-the-loop

Seretan® recommends; the investment committee decides. Every escalation carries a full, auditable evidence trail — why flagged, which signals, which model version, which disposition.

How it screens a Series A / B investor

Five steps. One committee decision.

From subscription documents to perpetual monitoring after close — Seretan® is built as a living control, not a one-time KYC file.

  1. 01

    Intake & entity resolution

    Subscription documents, cap-table entries and LP structures are ingested. Seretan® resolves every investor to its ultimate beneficial owners, unwinding vehicles, nominees and trusts across jurisdictions.

  2. 02

    Source-of-funds plausibility

    The LLM cross-reads declared wealth narratives against registry history, disclosed exits, tax-residency claims and public records — flagging inconsistencies a two-year-old vehicle cannot explain.

  3. 03

    Crime-risk signals

    Sanctions and PEP screening with AI adjudication of false hits; multilingual adverse-media analysis; network analytics linking the investor to known risk nodes — shared directors, addresses, wallets, counterparties. Typology detectors watch for round-tripping, structured sub-threshold stakes and late-round urgency.

  4. 04

    Risk memo & committee decision

    Seretan® drafts a plain-language risk memo with per-signal explanations and confidence levels. The investment committee accepts, conditions or declines. Dispositions feed back into model retraining.

  5. 05

    Perpetual monitoring

    Post-investment, Seretan® continues adverse-media, sanctions and network monitoring for the life of the holding. An investor who becomes risky after closing is detected — not remembered too late.

The capability stack

Eight capabilities. One integrity engine.

The same stack that production financial-crime programmes use at bank and FIU scale — applied here to the private-market context of venture-capital investor screening.

01

Intelligent monitoring

Behavioural baselining and adaptive thresholds on deal-flow patterns — not static rules that encode yesterday’s typology.

02

Network & graph analytics

Connects entities across vehicles, directors, addresses, wallets and counterparties. Exposes layering and shared nodes no per-investor file can see.

03

Entity resolution & KYC

Probabilistic matching, NER and multilingual name matching — including Arabic script — to one golden view of a person or company.

04

Sanctions & watchlist AI

ML name-matching with contextual adjudication. False hits close with auditable reasoning; recall on true matches is preserved.

05

Adverse-media intelligence

NLP over multilingual news, court records, registries and OSINT. Allegations, relationships and entities become structured risk signals.

06

Investigation copilot

RAG over case files and KYC. Assembles evidence packs and drafts the risk memo for human review — hours of narrative work, compressed.

07

Agentic workflow

Orchestrated agents gather, enrich and triage. Low-risk closures stay explainable; higher-risk cases always reach a named officer.

08

Virtual-asset & typology detectors

On-chain clustering where wallets appear; detectors for round-tripping, structured stakes and urgency patterns in late-round subscriptions.

Operating results inside UGF®

The control is the product.

These are results from UGF’s internal operations — not vendor case studies. They describe how the engine behaves on live deal flow. They are not a guarantee of outcomes in another organisation.

Hours, not days

Investor screening compressed from multi-day manual reviews to hours, with full evidence packs assembled automatically.

False hits, closed

The large majority of false-hit sanctions and PEP matches are auto-adjudicated with auditable reasoning.

One memo standard

Structured risk memos, consistent across rounds and geographies — committee quality does not depend on deal-flow volume.

Zero tolerated

No onboarding of sanctioned, PEP-undisclosed or source-of-funds-unexplained capital. Declines are documented for audit.

Seretan® metrics describe UGF’s internal operations. They should not be read as independently audited performance claims, and they do not transplant automatically onto a bank transaction-monitoring programme or a national FIU platform.

Operating principles

The rules Seretan® will not break.

AML-AI programmes fail for organisational reasons more often than technical ones. These ten principles are how Seretan® is run — and the pattern UGF offers to any institution building its own defences.

01
Start from risk, not technology

Use cases are anchored to actual typologies: opaque vehicles, PEPs, sanctions evasion, round-tripping, private-market legitimacy seeking.

02
Data foundation before models

Entity resolution, lineage and a governed analytics layer first. Models on fragmented data reproduce blind spots.

03
Human-in-the-loop by design

AI triages and recommends. Named humans decide anything with legal, reputational or capital consequences.

04
Explainability as a requirement

Every disposition is explainable: why flagged, why closed, which features, which model version.

05
Formal model governance

Inventory, validation, challenger models, bias and drift monitoring, governed change control.

06
Measure like a business

Operational KPIs (precision, closure time, throughput) and outcome KPIs (declines, conditions, committee quality).

07
Secure collaboration

Where sharing is required, legal and technical rails come first — never raw data into a public model.

08
Keep sensitive data sovereign

Investor files, model weights and audit logs remain on UGF-controlled infrastructure.

09
Build the talent pipeline

Analysts and committee members are trained to read, challenge and override the engine.

10
Design for adversarial adaptation

Continuous typology research, red-teaming and model refresh. A living defence, not a one-time delivery.

Who Seretan® is for

A private control. A public pattern.

Seretan® is the engine UGF runs on its own venture pipeline. The design principles — in-house models, entity-resolution-first pipelines, explainable memos, human gates, perpetual monitoring — are the same patterns a fund, a family office or a government financial-intelligence platform needs.

Funds & GPs

Protect the cap table you will have to live with.

For venture and growth managers taking external capital into Series A and B. Seretan® is how UGF treats investor integrity as a first-order control, not a side process.

  • UBO unwinding across vehicles, nominees and trusts
  • Source-of-funds narrative testing before IC
  • Sanctions, PEP and adverse-media with a human gate
  • Perpetual monitoring for the life of the holding
Governments & supervisors

From private tool to public good.

UGF does not sell Seretan® as a national FIU platform. We will brief institutions on the design principles, the governance regime, and how an in-house LLM is actually run — so a jurisdiction can build its own sovereign capability.

  • In-house LLM governance, not public-API dependence
  • Explainable dispositions suitable for supervisors
  • Arabic-first intelligence as a first-class requirement
  • A pattern, not a black-box product to import

Governance, trust & risk

An engine that touches capital must be more governable than the crime it fights.

No credible deployment removes humans from high-stakes decisions. Seretan® triages, accelerates and explains. People decide, and they are accountable.

Model risk & assurance

Inventory and lifecycle governance: development, independent validation, approval, monitoring, retirement. Drift and bias monitoring. Immutable logs of every score, feature contribution and human disposition.

Explainability & accountability

Per-signal plain-language explanations. Named officers decide committee-relevant outcomes. Automated closure is confined to validated low-risk classes — for example, demonstrable false-hit sanctions matches.

Data protection & sovereignty

Investor data never leaves UGF-controlled infrastructure for inference on a public LLM. Purpose limitation, retention schedules and need-to-know access. No training on your files by a third-party lab.

Adversarial resilience

Threat modelling against model evasion and data poisoning. Continuous red-teaming. Typology research as a funded cadence, not a side project after go-live.

Request a briefing

Sit with the team that runs it.

UGF welcomes briefings with funds, family offices, supervisors and financial-intelligence units. Tell us who you are; we will follow up from Dubai, Istanbul or Tallinn.

01
Send the requestRole, organisation and what you want to cover.
02
Scoping callWe confirm audience, confidentiality and whether a live walkthrough is appropriate.
03
BriefingDesign principles, governance, and — where relevant — how Seretan® behaves on UGF deal flow.

Required

Required

Enter a valid email

Required

Required

Request received.

Thank you. If your mail client opened, send the message to complete the request. Our team will follow up from Dubai.

Frequently asked

Questions, answered.

Seretan® is United Global Funding’s proprietary, in-house trained large-language-model engine. It is deployed across UGF’s Venture Capital arm to identify suspicious or high-risk investors seeking to stake capital in Series A and Series B rounds. It is not a public chatbot and not a third-party SaaS AML product.
No. Seretan® runs on UGF-controlled infrastructure. Investor data never traverses third-party public LLM services. That is a design constraint, not a setting.
No. Seretan® recommends; the investment committee decides. Automated closure is used only for demonstrably low-risk, fully explainable categories — typically false-hit sanctions and PEP matches with an audit trail.
No. Seretan® is UGF’s private-market investor-integrity engine. UGF’s global research report describes how the same design principles — entity resolution, graph analytics, in-house LLMs, human gates, sovereign hosting — apply to a government platform. That is a pattern to brief, not a product to install at an FIU.
Arabic, English and Turkish are first-class, reflecting UGF’s markets in Dubai, Istanbul and Tallinn, with further language coverage over public-source intelligence. Arabic-script name matching is treated as a core capability, not an add-on.
Briefings come first. Seretan® is the engine UGF operates on its own pipeline. Engagements with other institutions — if any — are scoped after a briefing, and would never involve sending investor files to a public model or relaxing the human-in-the-loop rule.

Global offices

Three regional centers. One platform.

Get in touch with United Global Funding® across the markets we operate in.