Investor KYC that unfolds every owner, screens every party and cites every page, so investors are accepted in days
Investors accepted in days, with every owner and hit cited to its source page.
See one case, screen by screen ↓Investor KYC/AML Onboarding is investor KYC and AML onboarding for asset managers, private markets firms and fund administrators: an analyst works each file in conversation with seven AI agents that collect the KYC documents, read registry extracts, trust deeds and IDs in any language, unfold the ownership chain to every ultimate beneficial owner, screen every party for sanctions, PEP status and adverse media, and rate money-laundering risk factor by factor, each statement cited to its source page. The analyst confirms, a second reviewer checks every file, the MLRO approves every high-risk investor, and the Refresh agent keeps every accepted file current.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
Every investor that subscribes to a fund — a pension plan, an insurer, a family holding company, a trust — has to be identified, its owners traced through holding companies and trusts to the natural persons behind them, every party screened against sanctions lists, PEP data and adverse media, and the investor rated for money-laundering risk before its money is accepted. High-risk investors need enhanced due diligence and the MLRO’s approval. Today a KYC analyst does it by hand: works out the checklist, emails the investor, chases each document, and reads registry extracts, trust deeds and IDs that arrive in Danish, French, Portuguese or Japanese.
The risk sits in the steps nobody can see: a percentage multiplied wrongly through a trust, a PEP family member missed because the link is in a foreign-language article, a screening hit discounted without a reason, a beneficial-owner register that disagrees with the trust deed, a periodic review that goes overdue. Meanwhile the fund closes on a date, and the rules keep tightening: the EU’s AML Regulation applies from July 10, 2027, and FinCEN’s AML rule for investment advisers is postponed to January 1, 2028 at the earliest.
Estimated split for a high-risk institutional investor with a holding company and a trust in its chain, by hand and with the solution.
Seven agents collect, read, unfold, check, screen, rate and refresh every file; an analyst confirms, a second checks, the MLRO signs.
Nordhavn Holding ApS, a Danish family holding company, is committing €40,000,000 to Quillfield Global Infrastructure Partners IV ahead of its first close on Friday, November 20, 2026. Its file, KYC-2026-04817, came in through the investor portal on October 6. It is Monday, October 19 — day 10 of a 15-day high-risk target. Here is how Leah Brandt works the file with the crew, screen by screen, in the working solution.
Leah owns 23 files in onboarding; 7 need her today, 3 could miss the November 20 close, and 46 screening hits are open, each with the agent’s proposal. Overnight the Ownership agent unfolded Nordhavn Holding, the Checker agent re-derived its 9 nodes on a second model, the Screening agent screened 342 parties across 61 files, and the Collector agent sent 31 approved reminders in 6 languages. Every statement carries a source chip, and the crew is one question away.
“I don’t mind the judgment calls. I mind the forty emails it takes to reach one.” — Leah Brandt
The file runs as seven steps — Collect, Read, Owners, Screen, Rate, Check, Approve — with the crew’s thread beside it. The Collector agent built a 12-item checklist for a private holding company in Denmark with a trust in its chain, and requested it in Danish and English once Leah approved the message. When Ana Lúcia’s proof of address came in older than three months, Leah rejected it with reason code C1; the Collector asked again in Danish and the new one came back on October 14.
The crew’s thread: “they propose, cite and ask · you decide”.
Nordhavn Holding ApS is 100% owned by Kestrel Bay Holdings S.à r.l. in Luxembourg, which is 62% owned by the Solvang Family Trust in Jersey and 38% by Mads Halvorsen. The trust deed gives Henrik Solvang and Ana Lúcia Solvang Moreira fixed halves, so each holds 31% — 62% × 50%. The trustee Halden Trustees Limited, the protector Lars Holm and the managing director Mette Kjær Lund are named as controllers. Each node carries its registry or deed text in the original language beside English, a screening chip and its risk factors.
As Leah confirms each node, the rating on the right builds: 44 with 5 of 9 confirmed.
The CVR registry extract — 3 pages from the company registry gateway, retrieved October 7 — sits in Danish beside the Reader agent’s English, with the passages it used highlighted. Each extracted field — legal name, registry number, address, legal owner, managing director, declared beneficial owners — is pinned to its page with its confidence, and Leah can correct any field with a reason code.
“Fields under 92% confidence go to the analyst before anything uses them.”
Ana Lúcia Solvang Moreira is a 31% beneficiary through the trust. The Reader and Ownership agents read 31%; the Checker agent re-derived it on a different model from the documents alone and agrees — and notes that the Danish beneficial-owner register leaves her out. The Ownership agent drafts the discrepancy report for the Danish Business Authority, for the MLRO to approve. Screening finds she is the spouse of Duarte Moreira, Portugal’s Secretary of State for Infrastructure: a PEP family member.
The Checker agent never edits a value: it shows both readings, and a disagreement blocks the node.
Gazeta do Estuário, March 14, 2026: prosecutors opened an inquiry into a port-works tender overseen by Duarte Moreira’s office; no one has been charged. The Screening agent shows the article in Portuguese beside English and proposes “relevant — allegation” (M2): it is about her spouse, it is an allegation, and it is in the fund’s own sector. A sanctions name match on “HALVORSEN, Mats” at 81% is proposed for discount (D3): a different date of birth, nationality and passport.
The Screening agent never clears a true match, and any match above 90% goes to a person.
Leah confirms the remaining nodes, each confirm adding its factors on the Quillfield Investor Risk Model v3.2: private investment holding company +12, three layers including a trust +10, Jersey trust +8, PEP family member +25, adverse media +9, the discounted sanctions match +0, among others. Every factor has a source chip and an edit pencil that needs a reason code. Any PEP among the owners sets High, so enhanced due diligence applies and the MLRO must approve.
“Risk agent proposes 78 · High · annual review · enhanced due diligence.”
Under the rating, Leah asks the crew “Why is it rated high?” The Risk agent answers: the PEP family member adds 25, adverse media about her spouse 9, the Jersey trust and three layers 18, the investor type, size, jurisdictions and holding layer 26 — each figure with a chip that opens its page, and a Show me that takes her there. She sends the file to the second check at 9:12 a.m.
“Any PEP among the owners sets High, so the MLRO must approve.” — Risk agent
Arjun re-performs the file against 14 checks the Checker agent pre-filled with their sources — owners of 25% or more verified, percentages re-computed through each layer, every hit decided with a reason, source of wealth evidenced, the register discrepancy handled — and passes all 14 at 10:05 a.m. Leah could not have: the analyst who prepares a file never checks it. Helena reads one page with the rating, owners and enhanced due diligence evidence, every line cited, and signs at 11:20 a.m. with reason code A2 — accept with enhanced due diligence.
“I sign for the investor. Give me the reason, the source and the risk on one page.” — Helena Marsh
Helena’s conditions: an annual review, next by October 19, 2027, adverse media monitored monthly, and any further commitment back to her. The crew prepares the transfer-agency record and Investor Operations releases the account as the second person — never the analyst. Every confirmation, answer, proposal and signature is on the audit trail with its evidence and the agent’s version, each entry sealed to the one before.
“Accepted in 10 business days, 32 days before the close.”
The same investor file, seen by five of the people who carry it — what their week looked like, and what it looks like now.
Seven agents collect, read, unfold, check, screen, rate and refresh every file; an analyst confirms, a second checks, the MLRO signs.
Builds the document checklist from the investor type, jurisdiction and risk, requests documents through the investor portal and email in the investor’s language, schedules reminders, and checks dates, certification and expiry.
Reads registry extracts, articles, trust deeds, IDs and letters in their own language, extracts parties, holdings, roles, dates and identifiers, and shows the original beside English with the line and page.
Unfolds the chain to every natural person, multiplies percentages through each layer, applies the 25% threshold (10% for high risk) and the control tests, names trust roles and the senior managing official, and drafts register discrepancy reports.
Re-derives every percentage and role on a different model from the documents alone, blocks a node when the readings differ, and re-checks every screening decision.
Runs sanctions screening, PEP screening and adverse media screening on every party — US (OFAC), EU, UK and UN sanctions lists, PEP data and news in any language — reads each article, and proposes a decision with its reasons.
Rates the investor on the firm’s risk model factor by factor, each factor cited, sets the review cycle and proposes enhanced due diligence measures.
Watches every accepted file for trigger events — registry changes, new designations, PEP appointments, adverse news, address or bank changes — opens periodic reviews on time and re-collects only what changed.
Maker, checker, signature, evidence. The agents propose; a named person decides.
Ask on the home screen, the file or the chain. The agent that did the work answers, every statement with a source chip that opens the passage, and a Show me that takes you to it.
Why is Nordhavn rated high?
Nordhavn rates 78 — High on Quillfield Investor Risk Model v3.2. The PEP family member adds 25, adverse media about her spouse 9, the Jersey trust and three layers 18, the investor type, size, jurisdictions and holding layer 26. Any PEP among the owners sets High, so the MLRO must approve.
Why does the register leave her out?
The Danish register lists Mads Halvorsen 38% and Henrik Solvang 62% through the trust, but the trust deed gives Ana Lúcia Solvang Moreira half of the trust’s interest, and the Luxembourg register already lists her at 31%. I drafted the discrepancy report for the Danish Business Authority; it goes to Helena Marsh to approve.
Show the Portuguese article
Gazeta do Estuário, March 14, 2026: prosecutors opened an inquiry into a port-works tender overseen by Duarte Moreira’s office; no one has been charged. It is about her spouse, it is an allegation, and it is in the fund’s own sector, so I propose “relevant — allegation” (M2). I opened it beside its English.
Which files could miss the Nov 20 close?
Three files for the November 20 first close could be late: Nordhavn Holding (with you, MLRO approval needed), Lindqvist & Berg Stiftelse (register and charter disagree) and Fundação Atlântica (two documents outstanding). I filtered the files to them.
13 screens from the working solution, on its sample data. Pick one to see it large.
Thresholds, cycles and the risk model are settings. An administrator proposes a change, the MLRO approves it, and both are on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Sanctions and screening lists | US (OFAC), EU, UK, UN · PEP data · adverse media in every language | Ongoing screening of every party on every accepted file, every night |
| Match threshold | 80% | At or above goes to an analyst; below, the agent proposes a discount with reasons |
| Sample of low-score discounts | 1 in 20 | Sent to a second reviewer |
| Former PEPs | 10 years | Treated as PEPs this long after leaving office |
| Ownership thresholds | 25% standard | 10% when the investor is high risk · trust roles always |
| Risk bands | Low below 35 · Medium 35–64 · High 65+ | Each factor weighted in the model; a PEP among the owners sets High |
| Review cycles | High every year · Medium 3 years · Low 5 years | Trigger events open a review in between |
| Targets | 5 business days from a complete package | 15 business days from receipt for high risk · escalate at 80% to the head of investor operations |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (a KYC software vendor’s September 2024 survey of 450 executives at asset managers with $51 billion or more under management, via Hubbis — $2,664 per institutional investor KYC review in the US and 74% having lost an investor to slow onboarding, vendor figures; a fund administrator’s figure via Blue Vault — about 20% of subscription packages complete on the first pass). Regulation (EU) 2024/1624 (AMLR) applies from July 10, 2027; FinCEN’s order of August 5, 2025 postponed its AML rule for investment advisers (31 CFR 1032) to January 1, 2028 at the earliest. Quillfield Capital, its funds, the investors, the newspaper and the people named on this page are fictional — characters and sample data in the working solution.
Investor KYC/AML Onboarding is investor KYC and AML onboarding for asset managers, private markets firms and fund administrators: an analyst works each file in conversation with seven AI agents that collect the KYC documents, read registry extracts, trust deeds and IDs in any language, unfold the ownership chain to every ultimate beneficial owner, screen every party for sanctions, PEP status and adverse media, and rate money-laundering risk factor by factor, each statement cited to its source page. The analyst confirms, a second reviewer checks every file, the MLRO approves every high-risk investor, and the Refresh agent keeps every accepted file current.
The Ownership agent unfolds the chain from the investor to every natural person, multiplying percentages through each layer — a 62% trust stake split in fixed halves gives each beneficiary 31%. It applies the 25% threshold (10% when the investor is high risk), always names trust roles and the senior managing official, and compares the result with the beneficial ownership register, drafting a discrepancy report where they differ. The Checker agent re-derives every node on a different model, and a disagreement blocks the node.
The Screening agent screens every party against US (OFAC), EU, UK and UN sanctions lists, PEP data and adverse media in every language, and proposes a decision for each hit with its reasons and a reason code. The analyst accepts it or decides differently with a reason. Matches below the 80% threshold can be accepted in bulk with one in twenty sent to a second reviewer; the agent never clears a true match, and true matches go to the MLRO.
Yes. The Reader agent reads registry extracts, articles of association, trust deeds, IDs and letters in their own language and shows the original beside English, each extracted field pinned to its page. In the working solution that includes Danish, French, Portuguese, Swedish, German and Japanese sources, and reminders go to investors in their own language beside English, for the analyst to approve.
The Risk agent scores the investor on your own risk model, factor by factor — investor type, jurisdictions, structure, PEP, adverse media, source of wealth, size — and cites each factor. The rating builds as the analyst confirms each owner, and any change to a factor needs a reason code. In the sample model Low is below 35, Medium 35 to 64 and High 65 and above, and any PEP among the owners sets High, which brings enhanced due diligence and the MLRO’s approval.
The Refresh agent watches every accepted file for trigger events — a registry change, a new PEP appointment, an ownership change, a sanctions list update, adverse media, a bank-detail change — and opens periodic KYC reviews on time: every year for high risk, every 3 years for medium and every 5 for low, the EU AMLR’s yearly and five-yearly minimums. It re-collects only what changed, and never re-attests a file or closes a trigger without a person.
Yes. The agents propose and cite; they never accept a document as verified, send a message a person has not approved, clear a true sanctions match, approve or decline an investor, file a register report or release an account. The analyst confirms, a second reviewer checks every file, the MLRO approves every high-risk investor with a reason code, and Investor Operations releases the account as the second person.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your risk model, document rules, screening rules and a sample of past investor files — and it goes live once every quality gate has passed. In the working solution, setup includes running 40 past files through the crew and comparing the results with the analysts’.
Every LP package in good order before the close, with the 25% plan limit kept live.
1-min demoFATCA/CRS Self-CertificationNo conflicting classification left on file, and one cited cure letter per investor.
1-min demoAccredited Investor VerificationVerified in minutes, every determination cited to the line it rests on.
1-min demoWe’ll run Investor KYC/AML Onboarding on a sample of your own past investor KYC files.