Company: Odeabank, Istanbul
Company Description: Odeabank established in March, 2012, to bring a new and dynamic breath to the banking sector, received the operating permit on October 2, 2012. Odeabank’s main shareholder, Bank Audi, headquartered in Lebanon and operating in 7 different countries. It operates in all lines of banking. As of the end of June 2020, consolidated assets of Bank Audi reached USD 37.1 billion.
Nomination Category: Solutions & Implementations Categories
Nomination Sub Category: Collaboration Solution – New or New Version
 Nomination Title: Generative AI Agent Assistant
- Which will you submit for your nomination in this category, a video of up to five (5) minutes, explaining the features, functions, benefits, and results to date of the nominated product or service, OR written answers to the questions? (Choose one):
Written answers to the questions
- If you are submitting a video of up to five (5) minutes in length, provide the URL of the video here, OR attach it to your entry via the "Add Attachments, Videos, or Links to This Entry" link above, through which you may also upload a copy of your video. If you are submitting written answers to the questions for this category, provide them in the spaces below.
- Describe the features, functions, and benefits of the nominated product or service (up to 350 words):
Total 188 words used.
The Generative AI Assistant project has been developed to improve operational processes, including advanced document processing, machine learning models, and automation technologies. The work has been completed by combining proven technologies with current technologies. The project aims to facilitate document searches through the guidance of RMs (Relationship Managers) and agents in contact with the customer, assist new hires in accessing policies, and increase the use of procedures. Before the development, new hires needed to be personally supported through the buddy system, but with the creation of the GEN AI Assistant, responses to procedures are obtained more effectively, and issues related to customer interactions are addressed more quickly. This solution has shown benefits in usage across both customer relations and operations departments.
Within the bank, 50 concurrent employees can actively query over 200 procedures. As procedures are updated, the latest documents are retrieved via APIs to provide the service. The responses given are reviewed by premium users, and feedback is collected. A system that evolves by feeding itself is planned to be developed.
The assistant not only handles document responses but also includes chat functionality and tool usage features.
- Reference any attachments of supporting materials throughout this nomination and how they provide evidence of the claims you have made in this nomination (up to 250 words):
Total 121 words used.
The feedback from user tests conducted on the Generative AI Assistant is considered a success for the project. User tests were specifically designed to challenge the Assistant, and support was gathered from users to push its limits. Questions were rated and recorded according to their level of difficulty.
Users reported a percentage decrease in the time taken to find the correct document related to their problems. The response time varies between 3 to 5 seconds. Answers to the questions were provided with an accuracy rate of 85%. It is expected that this success will increase over time with continuous feedback.
Key points highlighted in user comments include the ability to combine answers with additional information and reference-based blending of supplementary details.
- If you are nominating a brand-new product or service, state the date on which it was released. If you are nominating a new version of an existing product or service, state the date on which the update was released:
The Assisstant production launched on July, 2024.
- Outline the market performance, critical reception, and customer satisfaction with the product or service to date. State monetary or unit sales figures to date, if possible, and how they compare to expectations or past performance. Provide links to laudatory product or service reviews. Include some customer testimonials, if applicable (up to 350 words):
Total 231 words used.
During the planning of the Generative AI Assistant project, evaluations were made in various areas such as infrastructure-hardware limitations, data security restrictions, and document diversity. Considering these constraints, on-prem development is required because the data cannot be processed in a cloud environment, which leads to the need for a costly infrastructure. However, in the developed method, hardware costs were optimized by using the RAG architecture. As a result, over 200 documents were included without the need for an extensive training process. In the later stages of the process, document changes or updates will not cause any disruption in the flow, ensuring uninterrupted operation.
The assistant’s performance met expectations with an accuracy rate of 85.5% in user tests, demonstrating strong adaptability and performance. It has been effective in simplifying workflows, optimizing resources, and increasing efficiency.
In summary, the integration of the AI Assistant into business processes has led to success. The assistant has proven its ability to automate tasks such as document retrieval, content generation, and employee support. The AI assistant has significantly reduced operational costs and time, providing a seamless user experience for employees. Another unique feature is the creation of a "previously asked questions" database using the answers provided to questions. This database ensures that if the same questions are asked again, answers are provided without needing to query LLM services, while simultaneously feeding the learning process for training.
| Attachments/Videos/Links: |
|---|
|
