CaixaBank operates at the forefront of innovation - and was the first organization in the world to train the IBM Watson platform to understand Spanish when it built a decision-support solution for its traders. To enhance its customer service, the bank has again deployed Watson solutions to create a virtual assistant to support its client-facing operatives.
“Thanks to our work with IBM Watson, we are strengthening our position as a technology leader within the Spanish banking sector and maintaining our strong market position.” - David Almendros, Director of Artificial Intelligence, CaixaBank, S.A.
Business challenge
CaixaBank, S.A. employees need rapid access to product information when interacting with clients, but often had to rely on a call center for extra support. How could the bank share the information faster?
Transformation
CaixaBank used IBM Watson Discovery and IBM Watson Assistant solutions to create an AI platform that can answer most employee inquiries about products, services and financial regulations instantly.
Results
80% of queries answered instantly, reducing call center volumes
Accelerates access to product and service information for employees
Enables faster and more responsive customer services
Solution components
Watson Assistant
Watson Discovery
Business challenge story
Leading the way in innovation
One of the “big four” Spanish banks, CaixaBank has cemented its market-leading position through a decades-long commitment to innovation, helping it to deliver first-class customer service. In recent years, the bank has launched the first mobile bank in Spain, the first European implementation of contactless payment systems, and has pioneered the use of big data in strategic decision-making.
This visionary strategy has helped CaixaBank win many awards from international publications such as Global Finance Magazine, including “Best Digital Bank in Western Europe,” “Best Digital Bank in Spain” and “Best Bank in the World for Social Media.”
At the heart of CaixaBank’s customer service is a highly skilled client-facing team with a deep knowledge of its commercial banking products. If a client or prospect needs more detailed information, the bank’s client-facing teams can contact its internal call center, which receives around 500,000 inquiries every month.
To reduce pressure on its call center and accelerate the resolution of inquiries, the bank published more than 4,800 FAQs on its intranet, along with documentation on financial regulations. To enhance the responsiveness of its customer services, the bank wanted to make this valuable information faster and easier for employees to find.
David Almendros, Director of Artificial Intelligence at CaixaBank, explains: “We are constantly aiming to increase the efficiency of our internal processes to ensure our client-facing teams have speedy access to the information they need. Our call center and intranet were effective resources, but we saw an opportunity to provide our staff with even greater support.”
Transformation
Creating a virtual assistant for employees
Already a pioneer of AI in the financial sector, CaixaBank decided to build a virtual assistant to answer employee queries. The new platform is based on Watson Discovery and Watson Assistant solutions.
CaixaBank has led the way in introducing AI tools to the financial services. The bank was the first company in the world to train the Watson platform to read Spanish, and it used the solution to create a decision-support system for its foreign trading desks.
Almendros comments: “We have a long history working with IBM and have developed cutting-edge solutions together in the past. When we came to select a partner for our virtual assistant project, IBM was the natural choice.”
CaixaBank employees can access the virtual assistant via desktop or mobile devices and type their questions in natural language. The Watson solution then surfaces relevant information on products and services, internal regulations, financial legislation and more. If the chatbot cannot find an appropriate answer, the solution automatically starts a text chat with a member of the company’s internal support team. If the employee requires further support, he or she can transfer the conversation to the call center.
During the design stages of the project, CaixaBank worked with IBM to train the system with over 60,000 questions. As the virtual assistant receives more enquiries, its understanding will increase and provide even more valuable insights to the bank’s staff — and CaixaBank now has a dedicated team teaching the assistant to answer new questions.
Results story
Delivering superior client services
CaixaBank has now introduced the virtual assistant to all employees across its network of branches, which is the largest within Spain. Currently, the bank estimates that the chatbot can resolve 80 percent of all employee inquiries. By the end of 2018, CaixaBank believes the cutting-edge assistant will have answered 10 million different queries.
Almendros explains: “Our virtual assistant is set to have a major impact. As our staff can access the chatbot at any time and via their smartphones and tablet devices, they have all the resources they need to respond quickly to client questions and deliver an exceptional service. And because staff will no longer have to rely on making an enquiry via phone, we believe the project will significantly reduce the burden on our call center.”
He concludes: “Working with IBM continues to be an excellent and highly rewarding experience. IBM provides genuine expertise in the machine learning and AI space, and consistently delivers the tools we need to optimize our services and sharpen our competitive edge. Thanks to our work with IBM Watson, we are strengthening our position as a technology leader within the Spanish banking sector and maintaining our strong position in the market.”
About CaixaBank, S.A.
CaixaBank is a Spanish financial services company headquartered in Valencia, Spain. The company consists of the universal banking and insurance activities of the La Caixa group, along with the group’s stakes in the oil and gas and telecommunications markets, as well as maintaining holds in several other financial institutions. It is Spain’s third-largest lender by market value and has the most extensive branch network in the Spanish market.
About Incede.ai
Incede.ai is the AI division of Locus Solutions, Inc.; harnessing the power of Data, AI and Machine Learning
With more than half of the world’s data in natural language, Incede.ai solves business problems and creates a competitive advantage by leveraging the power of Watson AI and machine learning.As thought leaders, we collaborate with stakeholders to deploy AI solutions that deliver real value.
Roman’s Pizza
Chat your way into delicious food, satisfied employees and happy customers
Overwhelmed by high volumes of customer queries, Roman’s Pizza joined with IBM Business Partner Ocular Technologies to launched Levi, a customer support chatbot. The solution, powered by IBM Watson® technology, allows consumers to more easily find the food they crave while helping customer service staff stay a little saner.
“Where it used to take an average six to eight hours to hear back from us, customers now get a response immediately.”
— Matthew Jackson, Head of Digital Media, BPD Advertising
Business challenge
The customer service team at Roman’s Pizza was struggling to keep pace with the constant stream of questions being posted to its social media platforms.
Transformation
The business deployed a lateral conversation interface to handle common requests, such as those focused on store hours or weekly specials.
Results
90% of incoming queriesare answered automatically without human intervention
Responds immediatelyto customers, while previous replies took 6 – 8 hours on average
Frees up staffto focus on more important tasks and establish a healthy work-life balance
Solution components
● API Connect
● Cloud & Data Platform
● Watson Assistant
● Watson Tone Analyzer
Business challenge story
“It was a bit chaotic”
Pizza is delicious. Round flattened dough, rich tomato sauce, cheese and, depending on your taste, a mixture of meats and vegetables — there are few meals that are as universally appealing. And in South Africa, Roman’s Pizza tries to keep up with the ever-increasing demand for this treat, offering tasty, low-cost pizza through hundreds of restaurants spread across the country.
However, alongside this high demand for its pizzas, the business was also experiencing a high demand of support from its customers.
“We’ve got a fairly small customer support team,” explains Bonnie Cooper, Chief Executive Officer (CEO) at Roman’s Pizza. “So we weren’t able to keep up with questions and complaints in a timely fashion. We were getting inbound communications from customers 24x7 — long after our stores had closed — meaning some of these questions weren’t answered until the next morning, several hours later.”
These delayed responses, in turn, resulted in lower review scores on social media platforms, weakening the brand. In reaction, the customer support team began working longer hours.
“It was a bit chaotic,” recalls Matthew Jackson, Head of Digital Media at BPD Advertising, the marketing agency for Roman’s Pizza. “We were seeing all of these queries, and we didn’t have a real system in place to manage, escalate or log them. At one point, we had everyone from the management team involved. Even Bonnie was responding to Facebook questions from her home on the weekends.”
He continues, adding: “But many of these questions didn’t need to be answered by a person. We were having a lot of what we call Tier 1 queries — sort of mundane things. Where is my closest store? What are your specials? Are you open? Things that customers could typically figure out with an online search, but they preferred to ask someone.”
Aware of the recent advances in AI and chatbot technology and after viewing the technology demonstrated at the Lisbon Web Summit, Roman’s Pizza decided that it wanted to deploy a customer-facing communication platform that could resolve the majority of these requests without human intervention. Of course, the solution would also need to be sophisticated enough to know when a conversation needed to be escalated to human staff — avoiding further frustration for already upset consumers.
Transformation story
I’m Levi, your pizza ambassador
“Where it used to take an average six to eight hours to hear back from us, customers now get a response immediately.”
— Matthew Jackson, Head of Digital Media, BPD Advertising
Roman’s Pizza reached out to IBM, which quickly put the firm in touch with IBM Business Partner Ocular. A and the businesses developed and launched a lateral conversational interface called “Levi”.
“When we had a look at other bot platforms that were available, they were very linear,” explains Jackson. “Ocular presented a better proposal. They were the only company at the time that believed that they could achieve a lateral conversational interface that came across as natural language as opposed to a button-driven approach.”
Levi now acts as one of the company’s primary touchpoints for customer service, answering the majority of Tier 1 queries and helping customers to more easily get their pizza. The solution is currently available through Roman’s Pizza’s Facebook page, the company’s primary mechanism for communicating with customers, and will soon also be available on the restaurant chain’s website.
“We designed Levi to be anamorphic — in other words, very human-like,” notes Pommie Lutchman, CEO at Ocular. “It’s colloquial in terms of language. Humorous in terms of personality. Intuitive, meaning it has the ability to understand the tone and sentiment of the customer and respond in a natural manner.”
“I’ve got a love-hate relationship with technology,” adds Cooper. “I love what it can do, but it can be incredibly frustrating when it doesn’t work — I don’t have the patience to deal with poor user experiences. From the beginning, I wanted Levi to come across as a friendly customer support person, not a rigid bot.”
“We’ve even had some funny scenarios where people have thought that Levi was a real person,” Jackson adds. “In one case, we had a customer who thought that there was someone on the other end of the chat window that she could flirt with to get free pizza. Levi created a space where that customer felt very comfortable in who she was talking to — that’s when you know you’ve built a good conversational interface.”
To bring Levi to life, Ocular the solution utilizes arelies on its owncustom-built conversational interface, powered by IBM technology under an IBM Embedded Solutions Agreement (ESA). IBM Watson Assistant serves as the heart of the solution, while IBM Watson Tone Analyzer monitors chat instances, identifying when conversations need to be escalated to human staff. “It tracks the tone,” explains Lutchman. “And when it hits a threshold — swear words, aggression — it will send an SMS message to the support team that they have someone who needs help.”
In addition, the solution uses IBM API Connect® to integrate Levi with the Facebook Messenger app as well as IBM Cloud®, where the entire Levi infrastructure is hosted.
Results story
Trust the bot
On average, the Levi platform resolves over 90% of incoming issues and queries appropriately without human intervention, which dramatically reduces the demands on human staff at Roman’s Pizza. “In the past, we’d sit there on our cell phones and our computers and sacrifice family time to try and be as efficient as we can possibly be for our customers,” adds Jackson. “Now, we can give our team a healthier work-life balance, and they can explore the more exciting avenues of their careers.”
“Roman’s Pizza’s customer service response times have also improved,” notes Jackson. “Where it used to take an average six to eight hours to hear back from us, customers now get a response immediately. And if you have an issue outside of business hours that Levi can’t handle, it will log that complaint and notify the team to deal with it first thing in the morning. It addresses consumer’s concern up front and gives an expectation on when things can be resolved.”
Roman’s Pizza is also impressed with the capabilities of the underlying Watson™ technology that powers the new lateral conversation interface. “It was the tone and sentiment analysis that impressed us,” recalls Cooper. “Tone and sentiment are difficult even for humans. The fact that Watson could identify negative sentiments accurately was a strong selling point. We also liked the different languages that Watson could communicate in. While Levi is not scripted in multiple languages at this point, we have seen it have conversations with people who spoke other languages — Afrikaans speakers, for instance — and give answers appropriately.”
“I’m a big fan of IBM Cloud,” adds Lutchman. “There are so many services available — that excites us from an application development perspective. We can pull from all of these technologies to create complex solutions, and we do so in a reliable environment.”
“This project is not finished,” continues Jackson. “We are really excited to develop Levi out to its full potential. To put it into more places and to use it to help us solve more problems that we encounter. And that’s what we enjoy working with the Ocular team. They have been very attentive to our business needs and where we would like to see this project go.”
About Roman’s Pizza
Roman’s Pizza (external link) is a restaurant chain that is well known for its two-for-one pizza promotions. Originally founded in 1993 as Little Caesar’s, the business rebranded to the Roman’s Pizza name in 2002. It is currently headquartered in Johannesburg, South Africa and maintains over 230 restaurant locations, spread throughout the country.
About Ocular Technologies
Founded in 2003, IBM Business Partner Ocular (external link) is an IT solutions and services company that is headquartered in Sandton, South Africa. Taking advantage of the convergence of voice and data, the business specializes in developing and deploying customer engagement and digital experience solutions.
About Incede.ai
Incede.ai is the AI division of Locus Solutions, Inc.; harnessing the power of Data, AI and Machine Learning
With more than half of the world’s data in natural language, Incede.ai solves business problems and create a competitive advantage by leveraging the power of Watson AI and machine learning.As thought leaders, we collaborate with stakeholders to deploy AI solutions that deliver real value.