Мы используем файлы cookie для быстрой и удобной работы сайта. Выберите, какие файлы cookie вы разрешаете нам использовать. Подробнее в Политике конфиденциальности.
Обсудить проект
Отправьте запрос, и наши специалисты свяжутся с вами в течение 1 часа
Нажимая кнопку «Отправить», Вы даете однозначное согласие на обработку своих персональных данных в объеме и с целями, определенными Политикой в отношении обработки персональных данных.
Smart chatbot to help you choose the best bank deposit
E-commerce
AI-powered chatbots and the LLM
10
70%
the speed of request processing has increased
by
compiled the accuracy of the answers
time
Customer
A major Russian bank
CHALLENGES/FEATURES
High demands on the accuracy of responses due to image risks for the company
LLM's tendency to generate incorrect information (hallucinations)
Difficulty in finding relevant information in an extensive knowledge base
Routing complex user requests
LLM's tendency to generate incorrect information (hallucinations)
Difficulty in finding relevant information in an extensive knowledge base
Routing complex user requests
Developing a chatbot to help users choose the optimal deposit. The bot should provide information on available deposits at the bank, ask clarifying questions to the user to understand the context, assist in choosing the most appropriate deposit for a specific situation (term, amount, etc.), and be able to quickly respond to changes in the list or parameters of deposits by downloading an updated list of deposits as a PDF document.
Task
solution
Technical solution
Для реализации чат-бота, помогающего пользователям выбирать оптимальный банковский вклад, было принято следующее технологическое решение:
1. Dialog Platform:
The chatbot is built on the Aiogram library for interaction via the Telegram messenger, maintaining the dialogue context.
2. PDF Processing:
Current deposit data is loaded from PDF documents using the Python-docx library.
3. Server-Side:
The server logic is written in Python using FastAPI for high-performance request processing and Uvicorn for server startup.
4. Data and Logic:
Deposit information is stored in JSON format, which simplifies data updates. Deposit selection is performed using NumPy and Pandas for parameter analysis.
5. Monitoring and Debugging:
The Logging module is used to record events and errors, as well as Typing for type checking in the code.
6. Interaction with External Services:
The Requests library is used to send HTTP requests to external APIs.
The chatbot is built on the Aiogram library for interaction via the Telegram messenger, maintaining the dialogue context.
2. PDF Processing:
Current deposit data is loaded from PDF documents using the Python-docx library.
3. Server-Side:
The server logic is written in Python using FastAPI for high-performance request processing and Uvicorn for server startup.
4. Data and Logic:
Deposit information is stored in JSON format, which simplifies data updates. Deposit selection is performed using NumPy and Pandas for parameter analysis.
5. Monitoring and Debugging:
The Logging module is used to record events and errors, as well as Typing for type checking in the code.
6. Interaction with External Services:
The Requests library is used to send HTTP requests to external APIs.
Result
Business values
Increased speed of work
The accuracy of answers is above 70% (for the prototype)
Automation
Reducing the workload on technical support by 3 times
10 times increase in query processing speed
5x reduction in errors in data products
24/7 support
By clicking the "Submit" button, you expressly consent to the processing of your personal data to the extent and for the purposes defined in the Personal Data Processing Policy.
Development of software
and Big Data solutions
and Big Data solutions
Send a request and our specialists will contact you within 1 hour.
Choose a convenient method of communication
You can attach three files up to 3 MB each. Formats: doc, docx, pdf, ppt, pptx
Сообщение об успешной отправке!