IPsoft serves more than 550 of the world’s leading brands by providing technology that helps improve business processes. Executives from four IPsoft clients detailed their experiences with IPsoft’s cognitive AI digital colleague Amelia at the Digital Workforce Summit 2018. During the conversation, each client touched on the tasks Amelia has accomplished, the impact she’s had on their employees, and the ways in which she’s revolutionized business processes.
Vodafone, Amelia, and The IT Service Desk
Karine Brunet, Director of Technology Shared Services at Vodafone, said the mobile services company began using Amelia two years ago in order to support employees by resolving IT issues. Amelia is now available in seven markets in English, German, and Spanish, and 58% of employees who’ve contacted the IT service desk have used Amelia. Of the 20,000 chats per month that Amelia handles, she’s able to complete 53% without human intervention.
“We believe that in the next few months we’re going to increase that level of autonomy, and we think we should be close to 65 to 70% in terms of autonomy,” said Brunet.
Vodafone has integrated Amelia with 22 of the company’s back-end information systems, including its Active Directory and ITSM ticketing system, and more recently SAP SuccessFactors , the company’s HR system. Vodafone is leveraging Amelia in one use case to help employees find and register for training sessions. Employees tell Amelia what department they’re in, their roles, and which trainings they’ve already completed, and Amelia recommends new courses. She then automatically adds those courses to the employees’ training schedule.
Although employee reaction to Amelia has been good, Brunet said it’s important to Vodafone to let employees know they’re speaking with AI, just in case they’re not comfortable with the machine-to-human process. “We don’t want our employees to have the feeling they’re talking to a human when they are interacting with a bot, so that’s a very important principle for us,” she said.
Brunet said Vodafone is committed to its employees whose jobs might have been lessened as a result of Amelia. “We really look at AI as a way to improve the working environments of our employees,” she said. “But there’s also a commitment from Vodafone towards its employees that whether AI is misplacing potentially their jobs, we are repositioning and retraining those people to work on this newest technology.”
BBVA, Amelia, and Customer Service
Ignacio Bernal Echeverria, Global Chief Technology Officer at BBVA, needed to find a way to better serve the bank’s 17 million customers in more than 13 countries. With more than 8,000 branches, and more than 20,000 contact center operators, BBVA needed to resolve issues with large customer service volumes. In Mexico, just one of the company’s countries of operation, BBVA fields more than 100 million calls a year, so finding a solution that could tackle some of this workload was crucial.
“Volume is not a problem we expect to have in the future. It’s a problem we have now,” said Echeverria. “Lots of solutions promise to deliver in the future but we needed it solved now. So we needed some kind of relationship that was not a traditional IT-vendor, or IT-customer relationship. We needed some kind of a startup-like commitment, so that if we face some kind of challenge, we can work together to solve it.”
BBVA implemented Amelia to handle credit and debit card fraud claims at scale. “The solutions we need in the AI space need to be able to meet the requirements of a big company,” Echeverria said. “There are a lot of solutions for a startup, there are a lot of solutions if you are a small retailer. To handle [BBVA’s] volumes, there are not really a lot of solutions in the market.”
Today, Amelia manages 32% of the total calls for credit/debit card claims. “Imagine the impact for a customer that used to spend two, three, four weeks solving a critical dispute having those disputes automatically solved,” said Echevarria.
As a result of its success with Amelia, BBVA is looking into how to create “cognitive at scale” by introducing Amelia to other parts of the business. “How [can] we train Amelia in such way that the work we do for claims for the express segment will be reused for higher segments, or how the work we do for the commercial segments could be reused for private banking? So thinking on that cognitive architecture [level] rather than on individual cognitive challenges is one of the things we are starting to right now,” he said.
What about the contact center workers who previously handled calls that Amelia now handles? Echeverria said BBVA is still committed to those workers.
“We don’t feel comfortable at all to just eliminate all these positions by automating them through Amelia,” he said. “We certainly believe that in the communities where we operate, we need to find new ways of making sure that these labor workers could be reused in other things, or maybe we could improve their quality of life by reducing the amount of hours they work.”
Electronic Arts, Amelia, and Phishers
When Matt Tomlinson, Global Director of Innovation at Electronic Arts (EA), first deployed Amelia he was hoping for a solution that could reduce handle times for support. He didn’t realize AI could also be effective at reducing fraud.
“We naively thought early on that benefits were purely just handle time savings,” Tomlinson explained. “We thought Artificial Intelligence can do something in three minutes that humans can do in 10 minutes. Hooray. A seven-minute handle time savings. What we didn’t account for was that Artificial Intelligence is incredibly good at deflecting phishers.”
Amelia not only helped EA solve gaming issues for customers, she also thwarted efforts by phishers who were trying to (pardon the pun) game customer service agents into inadvertently handing over player credentials.
“Social Engineering can’t happen with Artificial Intelligence like it can with humans,” said Tomlinson. “So the amount of time that we are deflecting from phishers, real people were getting through to talk to our real live human agents, which was a phenomenal concept for us. For us, the ROI was actually an increase in handle time with the right people. We had better conversations with humans, to humans with the right customers. And then ultimately we found that real-time resolution and the post-engagement resolution was even higher with Artificial Intelligence than it was with humans.”
Tomlinson was also surprised by how naturally customers adjusted to conversing with Amelia.
“We thought our biggest challenge was going to be trying to make Amelia sound as human as possible. That wasn’t a problem at all,” he said. “Forty-five percent of our players had no idea Amelia wasn’t human. When they were transferred onto a human agent, they were asking, ‘Who was that Amelia person?’”
Faster Business Processes, Fewer Mistakes
For Mike Brady, former CTO at one of the world’s largest insurance companies, AI functions as a tool that simplifies and removes errors from standard business processes. “You can look at the business as a set of processes,” Brady said. “When those processes run well, sales get made, products get created and shipped, bills are issued, invoices paid and things are good.”
But, argued Brady, when processes can’t be completed, and multiple teams need to be looped in, business processes grind to a halt. When there’s an error, “[the] ingestion team has to send it to the sales department, which is in a different country. They go ahead, do the error corrections, they get the additional information and send it over to a quoting group. And the quoting group then goes ahead and has to look at product features, consult with product managers…But the problem we see is when you put all 14 of those steps together, suddenly your business process is a 14-day process.”
Brady said AI can help to remove some of the business process steps to streamline operations. “By looking at it differently, it opens up the application of Artificial Intelligence to replace our teams of middleware and connect those systems with the context, with the semantic, to complete the transactions.”
Brady said he’s seen transactions that took humans nine or 10 minutes to complete being done by AI systems in as little as seven minutes. “But what I didn’t expect is on average each one of those steps was sitting in a queue, sitting in someone’s work queue for three and half hours, and [with AI] the queuing delay was completely eliminated.”
He also said he’s seen improved quality thanks to digital colleagues: “We were dialing 97% [accuracy] very, very quickly, and in some cases we would run it very close to a 100%. A typical human on a good day is running about 96% accurate, and statistically there are times of the week that it’s actually getting a lot closer to 90%.”
More DWS Coverage
- Chetan Dube, IPsoft: The ROI on AI & Amelia Solutions
- Edwin van Bommel, IPsoft: Generating ROI for Your Business Through IPsoft’s Marketplace
- Anurag Harsh, IPsoft: The Business of AI
- Max Tegmark, MIT: The Acceleration of AI Is Transforming Business Practices
- Christopher Manning, Stanford University: Inside Amelia’s Brain
- Anthony Abbattista, Deloitte: The True Impact of AI
- John Brownridge, Deloitte, and Junichi Kudo, NTT Group: AI as the Way to Innovation
- Panel Discussion: Cognitive AI in Healthcare Means Augmented Patient and Caregiver Experiences
- Panel Discussion: Lessons Learned from AI Pioneers
- Panel Discussion: AI Deployment Surprises and Strategy