Eighty years later, its mission hasn’t changed: “To help people build better futures.”
“It’s core to everything we do,” says Irish Life’s head of information services, Ken Lynch. “Every time we look at systems or evaluate a solution, every time we consider a business problem, or examine an opportunity, we ask ourselves how we can help people build better futures.”
That commitment to customer service extends to the way Irish Life handles customer email.
Irish Life’s corporate division receives hundreds of thousands of emails every year, from both internal and external sources. Until recently, each was opened and read by a member of Irish Life’s email-triage team, who would index (classify) the email based on Key Business Indicators (KBIs) such as membership numbers or insurance-scheme identifiers, and forward it to the appropriate operations area. There the email would be opened and read again, and finally sent to an individual case manager for resolution. On average, each query took 2.5 days to resolve.
It wasn’t a hugely efficient process,” Lynch says. “We thought, if we could get a product to ingest those queries, understand what it is our customers want, and send them to the right area, we could serve these requests much more effectively.”
The trick was finding software that could do the job. It needed to be able to make sense of complex emails about insurance and pensions. It had to figure out what customers wanted, and route their email to the departments that could best help them. And it had to work without much guidance. “We wanted to be able to take raw data and drop it into the system, and have it understand what was going on,” Lynch says.
It was a tall order. To be honest, Lynch thought it might be impossible. Irish Life had looked into natural language processing years earlier and had been distinctly unimpressed. Every AI platform they reviewed in the past required a massive investment in IT and development.
“They promised a lot,” Lynch says. But when we dug into the details, we realized we’d have to write an awful lot of code to get the results we needed. That wasn’t the direction we wanted to go. We didn’t want to hire a lot of technical people to write a new codebase and revise it every time the software provider issued an upgrade. So we walked away from the whole idea.”
That was where they left it until Lynch met with Ushur on a trip to Silicon Valley. Ushur billed itself as a complete solution for intelligent automation, designed for industries with a high volume of customer contacts, like insurance, logistics and financial services. It offered conversational AI that could understand what customers needed and what to do next, along with a SmartMail™ feature for data extraction, which seemed perfectly suited to Irish Life’s KBIs. An intuitive visual builder and robust integrations with backend customer systems rounded out the proposition.
Lynch was intrigued but skeptical. “I questioned whether what they claimed they could do was even possible,” he recalls. But he took Ushur up on their offer of a six-week pilot.
Ushur’s customer success team quickly swung into action. It created a dedicated instance in a Dublin data center to comply with GDPR requirements, then worked with Irish Life’s IT to integrate Irish Life’s work management platform, Midas, using Midas’s native APIs. The integration enabled Irish Life to push incoming emails through Ushur’s workflow for KBI extraction and email classification. Ushur then inserted the extracted and classified values into Midas using a simple API call for each work item.
Lynch was impressed with Ushur’s agility and responsiveness—something he hadn’t expected in a company based thousands of miles away.
“We thought, ‘We’re here in Europe. How are the time differences going to work? How are we going to operate?’ But the team working with us has switched over to an Irish schedule, and we never miss a beat.”
“The other thing that's been truly impressive is how fast Ushur develops the product. As we gave feedback about things we'd like to see, the speed with which they were able to make changes was truly remarkable.”
Irish Life trained Ushur’s AI using hundreds of thousands of emails previously categorized by human reviewers (roughly 5,000 per category). To protect customer privacy and ensure compliance with GDPR, Ushur used its proprietary Data Anonymizer tool to remove all personally identifiable information (PII) from the emails before training and categorization. Irish Life then used Ushur’s SmartMail™ data extraction tool to create defined rules within a workflow to extract and return KBIs, which Ushur then populated in Midas using Irish Life’s APIs.
Right away, Ushur’s AI achieved 80% accuracy — comparable to human reviewers. And it only got better from there. The more email Ushur’s AI processed, the more accurate it became.
“The results we got were phenomenal, and we hadn't written a line of code,” Lynch says. “We’d never seen anyone who could do that. From our standpoint, Ushur were the only ones out there truly doing something different.”
With the success of the pilot, Irish Life deployed Ushur across its corporate division. Irish Life now auto-triages 70% of its incoming emails with Ushur and expects to reach 100% soon, at an accuracy rate of 95%.
Ushur improved Irish Life’s response time as well. Previously, it took Irish Life agents between 3 hours and 2.5 days to process enquiries and requests. Now, with Ushur, indexing takes less than a second and requires about 40% fewer resources. And since Ushur works 24/7, a backlog never piles up over weekends or holidays.
“The benefit you get from technology is that it reads and understands things in the same way every time,” Lynch says. “Whereas different humans can read and assess them in a slightly different way. Once you’ve trained the AI, it applies the same rules in the same way every time. It’s hard for a human to concentrate on menial tasks hour after hour. It’s not hard for the machine.”
Another benefit of automation is that Irish Life can reassign staff who previously worked triage to more complex, value-adding jobs, Lynch says.
“Our team are all intelligent, college-educated people. Our objective over the next few years is to automate their more mechanical tasks so they can focus on strategic delivery.”
The next step will be providing straight through processing on certain customer tasks by gathering data from customers using Ushur’s Invisible App™, a secure, 1-to-1 communication channel powered by conversational AI.
“Right now,” Lynch says, “if someone sends us a query, Ushur quickly indexes it and routes it to the right person within our organization. But that person might not open the query for eight hours. Only then might they realize they’re missing data they need to resolve the customer’s problem. So they write the customer an email, and maybe the customer doesn’t see it for a day. By the time they write back, another day has passed. That’s not only inefficient for us as a company, it’s a poor experience for the customer.”
“With Invisible App™, Ushur would understand what the customer needed and instantly respond with a request for the missing data. We could send that response in an email or a text, and the customer could then use Ushur’s Invisible App™ to quickly and efficiently respond to us. We believe we can remove a great deal of delay, both for ourselves and our customers, with the Invisible App™. It could be an incredible improvement.”
Ushur’s business value goes far beyond streamlining workflows, Lynch says. It’s inspired them to take a fresh look at the way they communicate with customers.
“Often in the insurance industry we direct people towards customer portals to resolve their queries. But what we found working with Ushur is that you need to think about how the customer wants to interact with you. Think of yourself, think of your mother, or one of your relatives, and how they would try to interact with big companies. What we’ve seen with Ushur is that it's all about customer ease.”
“Everyone has a smartphone now, whether they’re young or retired, and they feel comfortable using text-based channels of communication like SMS or WhatsApp. Working with Ushur is helping us rethink whether we should be pointing everyone to a portal or an app. Maybe it could be even simpler. Maybe you could simply communicate in plain text using Invisible App™, given that’s that’s how a lot of people want to interact with companies right now.”
“We've only started looking at that aspect of customer ease, and Ushur has been fantastic about bringing us to the start of that journey.”