Thousands of alerts a month, and a case file built by hand for each one
Dock receives 3,000 to 4,000 alerts a month, generated by monitoring customers' transactions and transactional behavior. This work is part of the Monitoring, Selection, Analysis and Reporting (MSAC) process, the model the Central Bank of Brazil sets out in its anti-money-laundering rules for regulated institutions.
Every selected case requires assessing the customer's registration, behavioral and transactional data, plus counterparty information where applicable. Analysts evaluate the evidence, the context of the transactions and the transactional behavior, and record their conclusions in a case file. When the analysis finds elements that amount to suspicion, this file underpins the report to COAF, Brazil's financial intelligence unit.
Before automation, analysts searched for, downloaded, extracted and organized all this information by hand. Each case took 3 to 4 hours, and most of that time went into assembling the file, not into the analysis itself.
Risk-matrix triage and case files assembled with artificial intelligence
Dock started using the VAAS risk management platform on two fronts.
A risk matrix as the triage layer. Built on the VAAS platform, the matrix works ahead of the risk assessment. Using account data, customer profiles and transactional behavior, it prioritizes cases showing atypical signals or other risk factors.
Case files assembled with artificial intelligence. VAAS applies AI to collect, consolidate and structure the information used in the risk assessment. The step of searching for and organizing data, previously manual, is now automated.
Evaluating the evidence and deciding whether to report to the authorities remain the responsibility of Dock's analysts. The technology handles the triage and the structuring of information.
Analysts used to search for, download, extract and organize information by hand to build the case file. Now that step is automated, and analysts focus on evaluating the evidence and the context of each case.
From 3–4 hours to about 10 minutes per case
The time needed to assess a suspicious-activity case and, when necessary, report it to COAF dropped from 3–4 hours to about 10 minutes. That is a 96% reduction, applied to a volume of 3,000 to 4,000 alerts a month.
The gain also changed where the team spends its time:
- Analysis instead of assembly. Collecting and organizing information is now handled by the technology. Analysts start each case with a structured file and spend their time evaluating the evidence and the context.
- Priority for what carries risk. The risk matrix orders the queue by atypical signals, so the most relevant cases come first.
- Human decision preserved. The decision to report to the authorities stays with Dock's analysts, as the Central Bank's AML model requires.
