RBS has built an artificial intelligence (AI) engine called Luvo to handle problems raised by bank staff without the need for a human advisor.
The bank believes that, if the trials are successful, the system could be used to handle direct customer queries as its intelligence and knowledge base increases.
Luvo has been running in pilots for a few months with 1,200 staff who handle queries from small businesses customers usually concerning problems such as lost corporate cards or forgotten Pins.
A customer would normally talk to the relationship manager who would then talk to an internal advisor about how to fix the problem before returning to the customer with the answer.
Luvo can do this far more quickly as RBS has taught the system how to respond to queries on these topics with the relevant information. Luvo has been able to deal with the problem itself around 30 percent of the time without requiring a human advisor.
Furthermore, the system can be accessed at any time of the day and should give uniform responses to the same queries each time to ensure that the correct information is provided.
Chris Popple, head of digital at RBS, told V3 that introducing Luvo to handle these types of queries frees up staff time to focus on more important problems facing customers.
“We’re trying to save our people’s time for more complex issues. If we can take all the questions around lost cards and have Luvo solve that, it can help our staff spend time on more complex conversations with customers,” he said.
Luvo currently has a knowledge base of around 20 key answers relating to simple service questions.
Popple said that the bank aims to expand Luvo’s capabilities to more complex areas as the system continues to improve.
“It could move to answer simple product questions like ‘Do I have travel insurance with my account?’ Then, over time, the hope is that it will not just answer questions but actually be able to take action to fix things. So you would tell it you’ve lost your card and it would understand and say: ‘Would you like me to send you a new one?’.”
The idea eventually is to put the tool into customers' hands, but Popple explained that there is a way to go before this can happen.
"We need to ensure it has enough knowledge and dialogue so it can understand and respond to customers accurately and with the right personality," he said.
RBS also stressed that it will not force staff or customers to use the system if they do not want to.
There is no set date for such a milestone, but Popple said that the bank is moving quickly. A second iteration of Luvo is set to be rolled out in a few weeks, and versions three and four are likely within the year.
The use of an AI system at banks is not new, as tools of this nature are often deployed to spot patterns and account anomalies for fraud detection.
However, RBS claims that Luvo is unique because it engages in direct real-time conversations and has been trained to recognise different accents and idioms from staff all over the country.
The system is also learning sentiment and emotion analysis to understand whether a customer is unhappy or frustrated and respond accordingly. This capability was not turned on during the pilot but could be in future updates.
Agile AI development
Popple explained that Luvo is being built in an agile way so that it can be updated as and when required while still learning in a live environment.
“It’s almost impossible to get something right in one go, and we’ve learned through the pilot that agile works really well and there’s no reason we have to go from a pilot to a full live product, so it’s effectively like a permanent beta,” he said.
Luvo is currently a text-based system run over an instant messaging interface akin to WhatsApp or Facebook Messengar (as pictured).
Popple said that RBS has the ability to allow voice interaction with Luvo, although at present the messaging interface seems the most sensible way of doing this as peope are comfortable engaging in this way, especially via mobile devices.
The tool has been developed with some core AI technology provided by industry partners. RBS did not reveal the names of these partners, but the 'human' element of Luvo has been developed entirely in-house.
Luvo is a key example of the new types of digital tool being embraced by businesses to improve everything from customer engagement to internal processes.
This long-sought progenitor star was identified in an image captured by Hubble in 2007
Intel's neural network USB stick could bring AI to the masses
Dubbed Barnard's star B, newly discovered planet is believed to be rocky
Also, what's a USB stick?