Visa Canada's AI chatbot walks customers through (non-payment) pain points
Visa Canada is collaborating with Finn AI to develop natural language chatbot technology to address common needs experienced by credit cardholders.
Issuers deploy conversational chatbots to act as digital assistants for their cardholders to free staff from routine inquiries about issues such as forgotten PINs, to answering more complex questions.
“We’re opening up the edges of our network through the Visa Developer Platform in order to improve the digital commerce experience for cardholders,” says Derek Colfer, Visa Canada’s head of digital product. “The actual payment represents a very small part of a consumer’s digital commerce experience. So, working with fintechs, we want to empower consumers in a way that goes well beyond just payments.”
The Visa Developer Platform provides fintechs, issuers and acquirers with access to around 100 Visa APIs in areas such as data and analytics; ID intelligence; offers and benefits; payment methods including the Visa Direct card-to-card transfer platform; and risk and fraud management. These APIs can then be used to develop digital services for customers.
Vancouver-based Finn AI began life in 2014 as the developer of Payso, a P2P money transfer app designed to be the Canadian equivalent of Venmo. However, the company realized after a year and a half of offering Payso that working with banks offered more opportunities than working directly with consumers. It then developed its chatbot technology for the banking and credit card industries.
With Finn AI's technology, possible interactions include notifying an issuer when cardholders travel abroad so their card isn’t declined at foreign retailers, and blocking lost or stolen cards. According to Jake Tyler, Finn AI’s CEO, around half of cards reported lost or stolen actually turn up within 48 hours.
Other capabilities that Finn AI and Visa Canada have developed include spending controls including limits on certain types of purchase, and the ability to check FX rates when using a card in a foreign store. This means a cardholder could decide whether to pay in their own currency or in the local currency of the merchant.
While Finn AI integrates with banks and issuers’ digital banking and card management apps, it also works with social media channels such as Facebook Messenger as well as SMS messages and Amazon Alexa.
Tyler estimates that Canadian consumers spend around 50 minutes a day using instant messaging apps. “They use instant messaging more than they use social media,” he says. “Messaging apps are increasingly becoming portals to the Internet and to digital services such as banking.”
Finn AI’s chatbot uses an AI technology called deep learning to analyze and learn from inquiries from customers.
“We think that deep learning is more flexible and adaptive than using rules-based AI technology,” says Tyler. “When a customer asks our chatbot a question it can’t answer, it hands over seamlessly to a human agent, but then researches the answer for the next time this question is asked.”
Other companies that have integrated with the Visa Developer Platform include Canada’s CIBC, which has developed a tool in its mobile banking app to give customers real-time foreign currency costs when making purchases abroad. San Francisco-based mass payout specialist Hyperwallet is using Visa Direct through the Visa Developer Platform in the Hyperwallet Transfer to Debit Card service.
Financial institutions that have deployed Finn AI’s chatbot include Banpro Grupo Promerica, Nicaragua's largest bank, and two Canadian FIs, Bank of Montreal (BMO) and ATB Financial (Alberta Treasury Branch). Commonwealth Bank of Australia plans to roll out the chatbot for its international banking operations outside Australia.
These implementations are separate from Finn AI’s work with Visa — BMO and ATB are both Mastercard issuers.
Finn AI’s core chatbot technology is card-scheme agnostic and can be used by issuers on any card network. Since around half of all Canadian retail purchases are paid with credit cards, Finn AI recognized that credit cards are an important use case for its chatbot, Tyler says.
“The Canadian credit card market is very competitive, with a lot of people switching between issuers,” Tyler says. “So the tools we have developed with Visa such as our travel notification capability can help build customer loyalty. If someone is traveling abroad and their card is declined, there’s a 44 percent chance that they will use that card less in the following 12 months.”
Finn AI already supports voice interactions with its chatbot, but its primary usage is with text-based messaging.
“I think conversational banking chatbots will be very powerful and deployed for many use cases,” says Colfer. “The advantage of Finn AI is that its chatbot can be used not just in a banking app but in a wide range of social messaging apps.”
Visa has a hands-off approach to the Visa Developer Platform, and doesn’t want to "curate" the development of new consumer uses cases by fintechs that join the platform, says Colfer.
In a separate development, Royal Bank of Canada (RBC) announced in March 2018 that it had become the first Canadian bank to launch an open API developers’ platform. Among the RBC APIs offered on the platform is a credit card catalog API, which provides access to credit card rates and fees, organized in terms of line of business, product family and individual products.
RBC’s APIs aren’t currently integrated with the Visa Developer Platform.
“One of the issues right now is the existence of a number of different sandboxes where developers are trying out APIs,” says Colfer. “So I think we will see platforms that aggregate multiple APIs from various sources in a universal sandbox to make it easier for developers to work with APIs from different businesses.”