NLP for Customer Services

How N26 Created a Banking Chatbot Cheaply

  • N26 wanted to build a tailored chatbot in a cost-efficient way. 
  • Open-source software (OSS) reduces the barriers of developing bespoke applications.
  • The bank used open source-software to quickly build its chatbot.
  • The chatbot now handles 20% of support requests in five different languages.

Chatbots have become a must-have channel

For banking services, chatbots often help with simple transactions, such as money transfers and balance inquiries. They reduce customer frustration, improve accessibility, and offer a consistent quality of services.

Among the many banks already deploying chatbot personal assistants are Wells Fargo, Capital One, Santander, HDFC and Hang Seng.

Leading the pack, though, is Bank of America, whose chatbot, Erica, already completed over 100 million client requests. Over a hundred people were needed to develop the application. This level of investment is only accessible to the largest financial institutions.

Open source natural language processing has opened up possibilities for smaller companies as well. N26, the mobile bank, has grown to over 3.5 million customers in just a few years. To improve customer experience, the bank decided to develop a chatbot, using open source software.

Concept: Open-Source Software

Under the closed-source software (or proprietary software) paradigm, the user pays for the right to use a piece of software, and intellectual rights, including the right to change the software, are reserved to its publisher.

This model implies that, to perform a task using software, users either have to pay for a license of an existing software product, or develop a bespoke one from scratch.

Software companies and independent developers are increasingly opting to provide Open-Source Software. Under this paradigm, users can use and customize the software for free.

Open-source licensing can limit commercial terms like re-selling or patenting. Open-source software has radically reduced the barriers to developing bespoke applications (like chatbots) for companies and individuals.

N26 used an open-source chatbot solution

When exploring potential chatbot solutions, N26 felt limited by the customization options offered by existing cloud-based solutions. Data-protection considerations were also a factor.

Using Rasa, an open source chatbot software, N26 was able to get from idea to production within four weeks. The bank’s chatbot, Neon, now handles 20% of all support requests through the bank’s mobile app.

The chatbot, called Neon, performs customer support tasks that range in complexity, from simple requests such as updating contact details to back-and-forth conversations, like reporting a lost or stolen card.

Crucially, as N26 operates in several European markets, the chatbot now provides services in five different languages.

How finance uses NLP