Our client is SGB-Bank—a bank that brings together the SGB Cooperative Banks, whose mission is to support local communities by providing modern and secure financial solutions. Operating under a partnership model, SGB-Bank develops technologies that meet customer needs and supports its member banks in achieving their business goals.
Ailleron has been collaborating with SGB-Bank for years, jointly developing the SGB Mobile app. The results of this partnership have been recognized by the jury of the Mobile Trends Awards for several years in a row, as evidenced by the company’s repeated nominations. One of the app’s main features is the Personal Finance Management (PFM) module, which uses artificial intelligence to automatically categorize transactions and help customers manage their household budgets more effectively.
Client
A bank that brings together SGB Cooperative Banks, which work to benefit local communities
Solution
Implementation of an AI-Based Transaction Classifier in the SGB Mobile App
Outcome
Automatic transaction categorization provides customers with greater transparency into their expenses and makes it easier to manage their budget
Project Objective
More Effective Transaction Classification
The client's goal was to increase user engagement with the SGB Mobile app by presenting transaction data in a more intuitive way and improving the classification of transactions made by individual customers.
Results of the Collaboration
The project delivered a number of short-term business benefits to the client
Greater user engagement and activity
The introduction of new features in the app led to increased customer engagement, a rise in the number of logins, and more frequent user activity.
Strengthening customer loyalty and better tailoring services to customer needs
By prioritizing user needs and providing more intuitive solutions, the bank builds greater trust and strengthens its relationships with customers.
Better Use of Transaction Data and Advanced Analytics
Structured spending data, obtained through automatic transaction classification, supports customer segmentation, product recommendations, data enrichment, and the creation of personalized notifications based on transaction history. Analyzing spending patterns also makes it possible to more accurately identify groups of customers with similar characteristics.
Results of the Collaboration
Business Workshops
We conducted a business workshop with the bank's team to develop and define the appropriate transaction categories.
Design of an Artificial Intelligence-Based Solution
We have designed an AI-based solution that enables the matching of customer transaction attributes to predefined categories.
Training a machine learning model
We defined the scope of the machine learning model training and developed a process for labeling anonymous data according to the established categories.
UI/UX Design
We developed a UI/UX design that integrates new personal financial management (PFM) features into the bank's existing mobile app.
At SGB Mobile, we believe that the best innovations arise from close, collaborative partnerships. That’s why, as we develop our services, we prioritize long-term relationships with technology providers, working together to create solutions that truly make our customers’ lives easier. A great example of this approach is the “Your Expenses and Receipts” service, in which we use the Expense Classifier provided by Ailleron. This is the first AI-powered solution we’ve implemented in our app. Our partnership involves not only implementing off-the-shelf solutions but also jointly shaping the future of digital banking. This allows us to respond more quickly to changing market expectations and deliver solutions that truly make a difference.
Artur Józefowski
Director of the Mobile and Online Banking Office, SGB-Bank
Scope of Collaboration
Step-by-Step Implementation of a Transaction Classifier
Integration with the bank's infrastructure
The machine learning model was developed as a microservice, enabling its deployment within the bank's infrastructure. The solution has been integrated with the banking systems responsible for processing customer transactions and with the mobile banking app.
Real-time classification
An event-driven architecture was implemented, which enables real-time transaction categorization. The Apache Kafka-based data streaming platform is responsible for collecting customer transactions and forwarding them to the transaction classifier.
Stable and secure architecture
The project utilized the non-relational MongoDB database and a suite of tools supporting monitoring, tracing, security, and log management, such as Grafana, Prometheus, Istio, Zipkin, and Fluentd.
Intuitive user experience
A UX/UI design was also developed, which enabled the integration of new personal financial management (PFM) features into the bank’s existing mobile app.
Analysis of Each Transaction
The classifier analyzes customer transactions, including bank transfers, card payments, online and mobile banking transactions, and local payment methods such as BLIK.
Automatic Categorization of Expenses
The classifier then automatically assigns customer transactions to one of the defined categories, such as: cars and transportation, cash, entertainment, hobbies and travel, education, daily expenses, health and beauty, clothing, savings and investments, income, and other incoming transfers.
Intelligent Analysis of Transaction Data
During the classification process, the tool analyzes various transaction attributes, including the amount spent, the transaction description or title, the recipient, the currency, and the Merchant Category Code (MCC).
High effectiveness of the model
The accuracy of the machine learning model powering the classifier is 92%, which means that 92% of customer transactions are correctly classified, whether they have an MCC code or not. At the same time, 98% of transaction values are assigned to the correct categories, which means that any errors are likely to occur mainly in transactions involving smaller amounts.
Better control over finances
As a result, individual customers gain a clear overview of the structure of their expenses and how it changes over time, as well as tips, alerts, and notifications to help them manage their budget more effectively.
