Noda Erudite Self-Learning Chatbot to Assist Promsvyazbank Customers

Promsvyazbank has recently completed the first phase of introducing a virtual assistant into its customer services. The chatbot which runs on the Noda Erudite platform will serve the bank's customers without involving live contact center agents. The bot can answer FAQs and render a number of standard banking services. It is planned to use the chatbot to answer customer calls both in text and voice channels.

In early 2019, the Noda Erudite virtual assistant will be available in test mode on the bank's website to a limited number of small and middle-sized business account owners. Then, by the end of Q1 2019, the chatbot will be accessible to any customer who wants to register a business and choose a bank service rate in accordance with the company's profile and business type. The chatbot will assist customers in opening accounts and advise them on the documents required by the bank; alternately, you can ask the bot to book a visit by the bank manager to complete the necessary formalities. Later on, it is planned to incorporate the virtual assistant into the bank’s customer service voice channels.

At the test-run stage, the chatbot is to process inbound customer calls in the small and middle-sized business segment. When fully operational, the virtual assistant will reduce the workload of the contact center live agents by 30 percent.

The virtual assistant runs on the Noda Erudite platform which is based on neural networks and machine learning. An important feature of Noda Erudite is its streamlined user interface that makes it easy to manage the full cycle of robotic services, e.g.: 

  • prepare training data (the assessor's interface); 
  • teach the bot subject-specific service scripts, including thematization, edit conversation scripts and answer templates, etc. 
  • teach the bot to respond to non-standard remarks (slang, counter and clarifying questions, interjections, greetings, expressions of discontent, etc.); 
  • monitor the bot's performance online and offline (dashboards, reports); 
  • analyze and process historical conversations to extract data for machine-learning. 
With the help of these tools, business users can train and retrain the bot without any help from IT experts or linguists.

The product allows you to upload large amounts of historical data that can be further sorted by topic. Using a special interface, assessors can select training data from the archive; analyze conversations, selecting the ones that can be used to train the bot, and divide them into separate lines; after that, the data can be loaded into the system and used for machine-learning.

Noda Omni-Channel functionality is used as an additional component of the solution. This allows to use unified scripts of robotized customer service in the bank's website chat and in its mobile app. Depending on the channel the call comes through, the robot can identify a customer on the basis of available data (if the mobile app is used), or request additional data in the conversation mode when the call comes via the website or by phone. One of the benefits of Noda Omni-Channel is that it helps to seamlessly switch between channels, both voice and text, combining them into a single ecosystem.

Andrey Zaitsev, Director of Contact Center and Robotic Systems Department at NODA: "At present, Noda Erudite bots serve over one million customer calls at a number of large-scale companies, such as Russian Post or the three big Russian electric power suppliers (Mosenergosbyt, Novosibirskenergosbyt, Omsk RTS). Our experience shows that robotization and AI projects involve much more than mere technologies. To a large degree, they rely on consulting. To create a robotized service script, you must carefully analyze the logic that contact center agents use. It may prove impossible to specify or formalize many of the servicing process elements. As a result, you will have to rebuild your current business processes. Cooperating with us, companies get both cutting-edge technologies and deep expertise, since NODA has been automating digital services for over 17 years now."

When selecting a chatbot platform, Promsvyazbank considered several developers and finally selected NODA for the technological maturity of its solution and its long-term experience as an integrator and expert advisor in the field of queueing optimization. Since 2001, NAUMEN has implemented more than 1,200 projects, many of them for large-scale companies, including Rostelecom, Inter RAO, Russian Post, The Rossiya Bank, OTP Bank, Sportmaster, Petrovich Trading Company, Jaguar Land Rover, etc. Due to its extensive project experience, the company has developed its own center of expertise and a powerful implementation team, which includes experts certified in compliance with ISO 18295 standard. This is a new standard in the field of customer service, developed by the International Organization for Standardization (ISO) operating in 163 countries.

Back to the list
Simply, conveniently, and quickly calculate the cost of Noda Contact Center licenses
Migration to the Noda Contact Center platform comes with a 60% discount on the licenses
We are currently looking for partners in the Asia-Pacific region