I just got out of a meeting at SRI International, where representatives from both SRI and international banking group BBVA showed off something they’ve been working on for the past couple of years.
Currently, SRI is best known as the research institute where Siri was developed before spinning out into a separate company and eventually being acquired by Apple, where it powers the Siri feature on the iPhone. SRI and BBVA have been collaborating on a new project, Lola, which they’re pitching as a successor of sorts to Siri. Bill Mark, SRI’s VP of Information and Computer Sciences, calls it “the next generation personal assistant”.
In this case, that personal assistant technology is being applied to a specific industry — banking. (Banking was, incidentally, one of the industries called out by Siri co-founder and former CEO Dag Kittlaus when he wrote a column for TechCrunch about the future potential of the virtual personal assistant.) BBVA Chief Innovation Officer Beatriz Lara Bartolomé says Lola is named after a real BBVA sales agent in Madrid, Spain. Her numbers were orders of magnitude better than anyone else’s, and when BBVA researchers interviewed her about the secrets of her success, they were most impressed by her passion and enthusiasm for the customers.
So with this project, they tried to create a virtual Lola.
“In a couple of years, you will love your bank,” Bartolomé promises.
In the demo, someone from SRI visited the website for BBVA Compass (the firm’s U.S. arm), where a chat window opened up. There was a “talk” button in the window, so the demonstrator hit it and asked out loud for the next payment amount for his loan. Lola answered. Then he asked for the payment due date, and he got an answer. He also set up a regular mortgage payment, and finally, he asked Lola to help him find a $14 transaction in his account — turns out he didn’t haven’t any transactions that were exactly $14, but Lola pointed out five that were close.
The tasks themselves may seem relatively straightforward, but Bartolomé says that the last one, at least, isn’t a feature that exists on banking websites — it’s probably something you ask for from a real human like a teller. Also impressive was the relative naturalness of the conversation. If you’ve ever interacted with a chatbot on a website, or gone through the automated menus on a customer service help line, you know that you usually have to stick to a very limited set of simple commands, phrased exactly right.
With Lola, on the other hand, the demonstrator was able to set up a complex task (making a monthly mortgage payment) with just one sentence. Lola also understood the context of different commands — when he asked for the balance of a loan, and then said, “Sorry, my mortgage,” the assistant realized that he was correcting an earlier statement and gave him his mortgage balance. When the demonstrator responded to a question (“How much would you like to transfer?”) with a question of his own (“How much do I owe?”), Lola didn’t bat a virtual eye — it answered his question, then brought him back to the earlier task.
Mark says that behind the scenes, Lola is using both understanding and reasoning — it needs to understand “what do those words mean in terms in terms of intent”, then it uses reasoning to determine the best way to fulfill that intent. In the future, Mark says Lola could start getting smarter and more proactive, for example warning you when you’re about to incur a fee or recommending changes that might improve your finances.
Over the next few months, Bartolomé says BBVA will be conducting field trials, with the goal of launching a commercial version of Lola “as soon as we can.” And while this would obviously be a boon to customer service, Bartolomé argues it’s more significant than that. She says that in a world where three-fourths of the population has a mobile phone, while less than 1 billion people have a bank account, Lola could become an important way to deliver financial services to the “unbanked.”
We might also see applications beyond BBVA. The company sees Lola as “quite a differentiator” for now, but it says licensing the technology to other banks is a possibility in the future. Norman Winarsky, SRI’s VP of Ventures, also says that SRI owns “the core engine” and would consider applying it to other industries.