-
Digital Trial Balance
Trial Balances (T-Balances) are free format financial documents that contain critical information of companies such as payables-receivables, deposits, long term loans and capital structure between companies and individuals.
Digital T-Balance is an artificial intelligence application, developed by using Natural Language Processing and Machine Learning technologies that can be used to extract and analyze information digitally and structurally from free- formatted t-balance documents. New customer engagement, up-sell & cross- sell and extraction of supply chain are some of the use cases.
Yapı Kredi used Digital T-balance services in the FOCA project in order to accelerate and increase the efficiency of credit evaluation and financial analysis process.
-
SAFIR – Smart Algorithms for Information Retrieval
SAFIR is an AI application that eliminates manual data entry by automatically processing customer orders transmitted to banks through different channels, in order to classify them and extracting the information in them.
Before using SAFIR, a large scale bank’s operation teams were responsible for determining the types of customer order transactions coming from various channels (e.g., fax and e-mail), entering their information into the system and verifying them. Thanks to SAFIR, all of these steps were automized and digitized. As a result, up to 50% reduction in manual data entry and 70% reduction in transaction durations were achieved.
SAFIR provides classification of 12 transaction types, including frequently encountered money transfers (EFT, money order, foreign money transfer), SGK payment, tax payment, and import payment orders, and extraction of more than 40 information-types (sender-recipient information, IBAN, amount, explanation, etc.) required for these transactions. The machine learning and deep learning models of SAFIR can be retrained with past transaction samples, if further coverage is necessary.
-
BOTYAP
Developed with no-code/ low-code approach, BOTYAP enables the creation of chatbots or virtual assistants for different sectors (banking, insurance, tourism, e-commerce, etc.) and different use cases (technical support, human resources, business processes, etc.) with quick integration.
By using a hybrid model of state of the art intent detection (deep learning, rule-based, dictionary usage, etc.) and current question-answer methods (answering within the document, similar questions), it ensures that different types of messages (short statements, long and detailed questions, etc.) from customers are understood.
Using natural language comprehension outputs (dialogue behavior, classification of intent, transaction information, etc.) and design preferences, it ensures that dialog flows can be dynamically created and managed. It also provides insights to improve the end-user experience by doing real-time analysis and clustering on past conversations.
-
ASSOS
ASSOS, a signature recognition and verification application, has the ability to determine the signature’s position, performing the presence check of the signature on the pages of free-form digital documents cleaning the stamps behind the signature, and estimate the similarity between two signatures with a percentage ratio. By specifying a lower limit on the similarity ratio, a dynamic binary decision such as “signatures are consistent/ signatures are not consistent” is also generated. The direction and color of the signatures being compared do not affect the performance of the similarity ratio.
It has been developed over completely different and free text bank internal documents such as instructions, contracts, and petitions, using Computer Vision algorithms and techniques in academic articles in the literature on signature recognition and verification in the last 5 years.
-
ROCKET
Rocket is an early warning system which predicts whether customers who get loans will have problems with their repayments in the future, and create early warnings. The forecasts are generated with machine learning techniques and can be from 3 to 9 months. With the help Explainable AI techniqus, the reasons behind the predictions can be viewed as well.
-
SİDE
SIDE is an application developed by examining the paying habits and common types of fraud. It enables the detection of credit card fraud by using machine learning methods. It adapts itself in a few days by automatically retraining its detection models, according to the newly developing fraud methods.
-
SİDE-C
In addition to ATM machines, telephone banking, internet and mobile branch banking channels are widely used by customers. Although these banking channels provide quick and low-cost transactions, they lead to an increase in fraudulent transactions.
Side-C is an application that allows the detection of fraudulent transactions implemented through different banking channels by using machine learning methods.
-
TAT
TAT is a text data analysis platform that provides insights into customers’ agendas by applying text analytics methods and unattended clustering algorithms on big text data. Thanks to this platform, customers’ all written and verbal communications from call center, complaint channel, social media, and chatbots, are examined instantly or collectively, making Customer 360 evaluation process possible.
-
CVPARSE
CVParse, used by the human resources departments, is an application that provides related information such as education and experience of applicants from their free format CVs by using natural language processing methods.
-
SAFİR-X
Safir-X provides efficiency in processes by enabling automatic separation and information extraction in business processes where documents are used extensively. Safir-X, which can be used in structured (ID, title deeds, etc.) and free format (instructions, petitions, court letters, etc.) document types, maximizes the targeted information extraction performance by utilizing both original deep learning algorithms and rule-based information extraction models.
By using Safir-X, which was developed with a no-code/low-code approach, the users can process the documents that they use in processes through configurations without coding and easily integrate them into their expert systems.
-
POS Cepte
By using “Yapı Kredi POS Cepte”, the users can turn their mobile phones into POS machines.
It is possible to create the “Yapı Kredi POS Cepte” application via Yapı Kredi POS Cepte, Yapı Kredi Mobil, Corporate Internet Branch, or from the nearest branch. Once the application is approved, the user can visit Google Play Store or Huawei Market Store to download the application The user can receive payment from contactless cards with the “Get Payment with NFC” option or “Get Payment with QR Code” through the NFC-enabled smartphone with Android operating system.
-
Mobil Onay
Without going to the branch, users can approve their documents related to product application or banking transactions initiated by the branches in accordance with the user request by using the application named “Yapı Kredi Mobil”. Copies of the approved documents will be sent to the verified email address or “Yapı Kredi Mobil- My Messages” menu. Target users of the mobile approval feature are, individual customers, customers who are qualified as commercial enterprises of real persons, and the legal customers who are limited liability companies.