From scattered investor data to a live AML onboarding and screening desk
A ~$600M discretionary manager in Vancouver runs equity and balanced mandates for high-net-worth and institutional clients. As a registered firm, it has to screen every investor against sanctions and watchlists at onboarding and keep that screening current. But the records behind it lived in OneNote and Excel, onboarding ran over email and phone, and screening was done by hand. We built the structured data layer first, then built the AML desk on top of it.
Vancouver, Canada·June 2026·6 min read
~75%less manual screening time per onboarding
Same-dayonboarding, down from 2 to 3 days
100%of clients re-screened on every list update
~1 hrrefresh cadence from government list sources
Figures are illustrative of the engagement and rounded; exact baselines vary by reporting period.
The challenge
Building a compliance desk on a firm that wasn't structured for one
The firm ran equity and balanced mandates for high-net-worth and institutional clients across a book approaching $600M. As a registered dealer, it had to screen every investor against federal sanctions lists, PEP registries, and criminal code watchlists at onboarding and keep that screening current. The records behind it lived in OneNote and spreadsheets, onboarding ran over email and phone, and screening was done by hand. As the book grew, manual checks became a constraint on how fast the firm could take on new clients, a single stale list change could leave the whole book out of date, and responding to an audit inquiry meant reconstructing what had been done from files that were never designed to answer those questions.
None of the automation was possible until the data was usable. We built the investor record first: a single structured source holding identity, account details, holdings, and transaction history, drawn from the systems the firm already used. That immediately unlocked automated XIRR reporting across six standard periods per investor. But reporting was the smaller return. A clean investor record is a platform, and once it was in place, the compliance layer had something real to build on.
"The rules don't care that we're small. The obligation is the same as everyone's, and we were meeting it by hand."
Principal, boutique Vancouver investment manager
The AML desk
Screening that runs itself, with a human on every decision
Built on: Client Onboarding & AML Automation
With the investor record in place, the compliance layer was straightforward to build on top of it. New investors now come in through a structured intake that captures identity, suitability, occupation, and PEP status as real fields, not free text in an email. Submitting the intake triggers an automatic screen against every watchlist a Canadian firm has to clear: federal sanctions and terrorism lists, Criminal Code listed entities, and politically exposed persons. Each source is pulled from its government feed and refreshed continuously, so the firm always screens against current data. Any change to a list re-screens the entire book, not just new clients. Matching uses name plus secondary identifiers like date of birth and nationality to control false positives, and anything above the threshold opens a case with a reference number, a confidence score, and the entry it matched.
Investigate · AML-2026-0187
Strong hitAndrei Volkov · DOB 1971-04-08SEMA Russia Consolidated
Match confidence87%
AI recommendation · Likely false positive
Name and DOB match exactly, but this client has lived in Canada for 12 years running a legitimate import/export business; the listed person is a senior foreign official. No adverse media. No STRs in 8 years of history.
system screening run triggered on onboard submit
agent drafted rationale & recommendation
analyst reviewed, dismissed as false positive
Each case carries a match score, the agent's written recommendation, and a complete log of every action taken. Sample data shown.
Triage is where the real work is. For each case, an AI agent reviews the match details, the listed individual, the client's history, and any adverse media, then writes a recommendation in plain language: clear, potential hit, likely false positive, or escalate. By the time a case reaches an analyst, it is already sorted and explained. The analyst reads the reasoning and makes the call. Nothing clears itself, and no match is dismissed without a person signing off. Every screening run, case, and analyst action is written to an append-only log on the investor record. When a regulator asks what was screened and why a match was dismissed, the answer is already there.
"It reads like notes from a junior analyst who never sleeps. I still make the call, but I start from a written rationale instead of a blank box and a screenshot."
Compliance lead, boutique Vancouver investment manager
Results
A compliance desk that runs on its own, on data built to last
~75% less manual screening time per onboarding, as scored matching against live government lists replaced one-off manual checks.
Same-day onboarding, down from 2 to 3 days.
The whole book stays screened, with every investor re-checked automatically whenever a list changes, not just at intake.
Every decision documented with actor, timestamp, and the agent's written rationale, so the audit trail builds itself.
The bigger result is harder to put in a single number. The investor record now holds performance, compliance history, and account context in one place. Whatever the firm builds next, whether periodic risk reviews, richer client reporting, or further automation, starts from that same foundation instead of from scratch. That is the real return on structuring the data. The AML desk was the first thing it made possible, not the last.
Elevate your impact with us.
Learn how we can help you reduce operational friction and make timely, defensible decisions.
