One Zero Bank Integrates Financial Data with AI Chatbots
The digital bank 'One Zero' is introducing a feature allowing customers to connect their financial data directly to AI agents. This move aims to provide real-time financial analysis within existing AI platforms without manual data entry.
The digital bank 'One Zero' will allow its customers to connect their financial information directly to artificial intelligence agents. This move is intended to allow users to receive analysis and insights into their financial situation within the AI platforms they are already using, and later to perform actions as well, as part of the bank's concept of becoming an adapted data infrastructure ('Embedded Finance') instead of remaining just a closed application.
As of now, the bank does not yet present this option in the application, and those who wish to gain access will have to contact the bank proactively. In the first stage, the service is available through ChatGPT only, and later an expansion to Claude is planned, and then to other tools such as Gemini.
Technological Implementation
Technologically, the connection is made using a dedicated connector component that the bank developed. Once the customer connects the connector to the AI platform in which they operate and identifies themselves securely in the bank's systems (OAuth protocol), a direct communication is established between the general AI agent and 'Ella', the financial agent of 'One Zero'.
The connector allows the external agent to pull processed and classified data in real-time from the bank's infrastructure layer, without sharing passwords or relying on uploading Excel files and incomplete screenshots. In the bank's view, the public tends to concentrate the management of their daily routine in large language models.
Consumer Habits and Rationale
The rationale behind the move is based on a deep change in consumption habits. According to the bank's data, about 62% of retail investors in Israel do not use a professional factor at all for making financial decisions, and 85% are not willing to pay for human advice. On the other hand, 58% of independent investors already rely on AI tools, and among young people up to age 35, the rate jumps to over 70%.
To avoid errors resulting from raw data, the bank does not transfer the information as is: the bank's digital agent, 'Ella', processes and classifies the movements and categories in advance, and presents the AI model with cleaned, processed information accompanied by general statistical benchmarks, which reduces the risk of 'hallucinations'.
Security and Future Steps
Regarding information security and privacy, the bank emphasizes that its responsibility applies to the information existing in its systems and its secure transfer. Once the data reaches external platforms, the transfer is carried out under the customer's responsibility and with their explicit consent, valid for a year. To maintain privacy, it is recommended to configure the chat being used so that personal data is not used to train the artificial intelligence models.
At the current stage, the connection is limited to viewing and analysis. In the next stage, planned for a few months from now, a layer of performing actions (Supervised Autonomy) will be added — where an agent will be able to prepare instructions for transferring funds, but the actual execution will always require active confirmation of the customer within the bank's application.
