Redesign the user experience of AML and compliance tools for fintech teams.
ROLE
Principal Product Designer
PERIOD
May - October, 2022
Flagright is a YC backed AML (anti money laundering) fintech startup that helps other fintechs simplify adapting compliance protocols and transaction security through their real time transaction monitoring API that allows fintech teams to flag and track suspicious transactions in real time.
I joined Flagright in May 2022 as their Principal Product Designer. I redesigned the Case Management and the Transaction Monitoring products to improve the user experience for compliance and fintech teams.
TRansaction Monitoring
Flagright provides a real time transaction monitoring API that fintech teams can use to monitor and flag all transactions taking place on the platform in real-time. They manage to do this by providing a set of tools that automate fraud detection and create case files for case officers automatically.
Fintech companies are required by law to monitor and report all suspicious transactions taking place on their platform in order to prevent any money laundering (ML) activities.
identifying Risky transactions
The first step to monitoring transactions is to determine the probability and level of risk that a transaction is fraudulent or part of a ML scheme.
For this, Flagright allows users to set risk levels for different parameters of a transaction - payment method, send and receiver nationality, age, etc.
Adjusting risk parameters
An example of adjusting risk is for geography or nationality of a sender or receiver. Users can select continents and countries and assign a risk level, either for a group of, or individual geographies.
manage risk automatically with rules
Rules allow users to automatically take action on transactions when certain transaction parameters cross pre-defined thresholds at different risk levels.
Defining rules
Rules allow users to automatically take action on transactions when certain transaction parameters cross pre-defined thresholds at different risk levels. These rules are applied per transaction.
First transaction
First transaction amount on new payment method is greater than (x)
Isolated crypto user
A user makes >= (x) crypto transactions without any fiat transactions.
Transaction amount too high
Transactions amount is USD (x) higher than the expected USD (y)
High volume sender
(x) number of transactions from the same sender in (y) seconds.
create and manage a RuleS Library
The rules library allows teams to use existing rules or create new rules against which each transaction passing through their system is compared in order to identify suspicious transactions.
Original design for the Rule Library table using the default ANT Design component system
Redesigned Rule Library that displays the name of the rule, description, and how the rule behaves for different risk levels. Public Rules and Private Rules are separated using tabs.
create rules with machine learning
Machine learning allows teams to automatically create transaction and user cases. Users set a confidence score threshold. If the score assigned to a transaction crosses the set threshold, a transaction or user case is automatically created.
To preview the impact of changing the confidence score, I added a Prediction section that provides a percentage approximation of the number of cases that would be created from all transactions.
I added Explanation and Prediction sections to explain how the number of cases created corresponds directly to the change in the confidence score.
Spiderview
Spiderview allows teams to track transactions from senders to receivers across multiple nodes and geographies within a single transaction.
Evolution of the Spiderview design
I added a fixed, interactive date and time range selector to the bottom of the Spiderview which displays all the transactions that have taken place within the selected period.
Additionally, I also designed a Transaction Bar that displays all the transactions and their statuses below the corresponding dates in the date range picker.
Case management
When a transaction has been flagged, suspended or blocked, the compliance team is usually interested in learning more about the sender and receiver, and particularly, their financial history and health.
Compliance teams create a case file that captures all the information about the transaction, sender and receiver. They can add additional notes, files, links and comments to this case file to give context as to why the transaction or user has been flagged.
TRANSACTION CASES
Transactions that hit one or more rules in the Rules Library automatically appear under the Transactions tab of Case Management. Compliance teams use transaction data displayed for each transaction to determine if the right action has been taken on a transaction.
With machine learning enabled, users define a confidence threshold above which the AI automatically updates and closes cases, allowing users to only investigate the most important transactions and cases.
Early designs of case management UI using the ANT Design Framework
Redesigned case management UI along with tabs to separate transaction and user cases
USER CASES
When a transaction case is created, a user case is automatically created for both the sender and the receiver. If the user case ID already exists, the existing case ID is updated with the latest transaction details.
In this way, compliance teams can compile a comprehensive dossier of a user's activity and if required, alert the police or the FATF (Financial Action Task Force).
Designs for all the different states for when users try to find a customer
USER CASE Details
Case details are meant to be a comprehensive knowledge base of a user's demographic, financial and transaction history on a financial platform.
Compliance teams use cases to document updates and feedback about a user to build a case and eventually use this information as evidence when closing a case and recommending next steps.
Customers wanted to be able to add comments and attachments to a case while scrolling through the case file. To accommodate this, I added a comments bar that stayed fixed to the bottom of the screen, so that users could add comments as they scrolled through a case.
Closing a case
Once a case is reviewed, the compliance team needs to close the case and recommend the next steps that need to be taken. When closing a case, the team needs to provide reasons and evidence for closing the case.
A closed case can always be re-opened by a team member if they feel that the case has not been sufficiently reviewed.
Case insights
We wanted to provide financial teams with quick and insightful data that they could take action on quickly, without having to manually browse and filter through transactions.
I designed a series of panels that represented important information in the form of graphs and charts. These panels could exist in either single, two or three column layouts.
Learnings
As a principal designer, Become Deeply familiar with the industry.
I was unfamiliar with fintech as a whole, and even more unfamiliar with compliance the the world of financial crime. While I tried to educate myself about the compliance industry, there was a lot of information that could only be learned through real world experience and learning from a network. And while I found my groove towards the end of my stint, I wish I had reduced my knowledge gaps earlier.
Grid layouts are great for unpredictable interfaces
We wanted users to be able to customize their dashboard and case management interface, since different users are interested in different types of data. I iterated with 3x3 and 4x4 grids to create resizable panels that users could move around the UI, depending on where they wanted to see the information.
Acknowledgements
I appreciate the opportunity given to me by the founders, Baran Ozkan and Madhu Nadig. I enjoyed working with them and the Flagright team and am happy to see them grow and succeed.
Frankensteined in Mumbai.
Fueled by YouTube, 90s cartoons & Heavy Metal.
