How can companies avoid AI fragmentation, with disparate tools and platforms operating in various parts of the business? That’s a scenario that MetLife, the Manhattan-based insurance giant, wanted to avoid.
“We started with the mindset that we needed a composite AI platform, not tools for the sake of tools,” says Bill Pappas, EVP and Head of Global Technology and Operations at $71 billion MetLife.
So in 2023, the company launched a unified AI platform called MetIQ to help teams quickly create and deploy new software apps; improve customer service by increasing the speed and quality of the interactions; and reduce workloads for employees. The platform also allows MetLife to plug in emerging technologies such as generative AI and agentic AI to help the company continue evolving its capabilities.
Pappas recently spoke with InnoLead about the development of MetIQ, and how his CEO regards AI as a “force multiplier” to the company’s strategy.
His Role
Bill Pappas: I joined MetLife right before the pandemic, in November 2019, after spending almost 20 years with Bank of America. What drove me to MetLife was its strategy. It sounds very simple, but the MetLife purpose is, “always with you, building a more confident future.” It puts our customers at the center of everything that we do, and that was really exciting for me from an opportunity perspective.
I manage what I call a consolidated technology and operations function, responsible for all disciplines that bring the customer experience to life every day. That function includes application development and infrastructure, data and analytics, cyber and physical security, call centers and operations, enterprise change and process engineering, and even real estate and crisis management.
My team manages close to 40,000 resources across more than 40 markets where we do business, serving 90 million customers.
The Project
BP: Six years ago, we asked, “How does technology add value? How does technology become a force multiplier? How does technology allow you to win in the marketplace?” …The first step for us was to modernize our infrastructure and simplify legacy systems before introducing any emerging technologies.
Then…as we thought this through, we said we would do it in a way that would be architecturally flexible, allowing us to build innovative solutions ourselves or collaborate with leading tech companies to deliver with greater speed, simplicity, and transparency. We developed a composite AI platform that enables us to leverage multiple AI techniques and rapidly incorporate emerging technologies like generative AI and agentic AI. This puts us in a position to continuously evolve our solutions.
We decided to name the platform MetIQ, because it reflected the platform’s ability to keep us informed as we navigated these changes, and it has become a force multiplier in moving forward. We started with the mindset that we needed a composite AI platform, not tools for the sake of tools. With MetIQ, we created an internal AI platform, built by our own engineers who have a deep understanding of our business and how our people actually work. From an architecture perspective, it connects our front end, which is our channel that includes chat, voice, APIs, and digital, with composite AI solutions from automation to generative AI to agentic AI, while ensuring architectural security and responsible use of AI are embedded throughout these capabilities. MetIQ allows us to scale our capabilities through these composite AI solutions and is architecturally flexible, so we can build ourselves or partner with leading tech companies.
…We developed a sandbox, which is a dedicated space that mirrors MetIQ’s capabilities, but is…away from production. That is very important for us, because it allows our associates to experiment in a secure place…
That allowed us to take the MetIQ platform and plug it into an environment that is becoming increasingly modern as we invest heavily in our infrastructure going forward.
We also needed a methodical way to provide solutions back to our associates, in a scalable way, that blended different interfaces with what we currently had available, while ensuring we had a clear understanding of both large and small LLMs and that it was secure for our customers and company assets.
In addition to making MetIQ available to our associates, over the last six months, we developed a sandbox, which is a dedicated space that mirrors MetIQ’s capabilities, but is…away from production. That is very important for us, because it allows our associates to experiment in a secure place, while making all the capabilities of MetIQ available to them. They can experiment with real business issues and not experiment for the sake of saying the tools are working or not. This kind of work was a very different mindset for us.
For us, AI is a force multiplier to our strategy. It’s not a separate strategy.
How It Got Green-Lit
BP: One of the things our CEO, Michel Khalaf, did was never look at AI as a technology imperative. He saw AI as a force multiplier for a strategy that everybody across the leadership table owns. That sounds very simple, but it’s very, very powerful, especially when you have to ensure the right alignment.
What we have done is align MetIQ back to those very specific business outcomes and prioritize the work through a commercial mindset that aligns with our strategy. For us, AI is a force multiplier to our strategy. It’s not a separate strategy.
The Three Biggest Challenges
BP: The first one was how do you actually architect something that is integrated into multiple forms of AI into what we call a single, secure, and scalable platform embedded in the cloud to enable a scalable, flexible infrastructure? It sounds very easy, but it’s a little daunting, because we wanted multiple things. We needed to combine the technical aspects of the platform with the responsible use of AI and understand what that really means. When you look at all the pieces we were putting together, like cybersecurity, the cloud, and AI capabilities, we needed to make sure we understood them and could use them at scale.
The second one was that we needed to keep our focus on our legacy system and its modernization. When you look at what’s available and what’s provided through MetIQ, you can only take full advantage of those capabilities if you prioritize modernization. Although we had done many of these modernization things before we implemented MetIQ at scale, we need to continue ensuring our environment stays contemporary.
The third challenge was making sure we understood that it’s one thing to create something cool from a technology perspective, but another to ensure it gets enterprise-wide adoption. For me, you have to ask “so what,” and why are you creating this? You can get very enamored about the architecture, your legacy, and how to be very contemporary. And it’s an expensive proposition. For us, we spend a lot of time making sure we understand the why and creating a portfolio that really balances the opportunities from a process perspective, a key activities perspective, and a tasks perspective. In understanding the processes, we needed to rethink how we serve customers from our call centers. We needed to rethink how we view claims. We needed to rethink the way we’re developing new capabilities and introducing products into the environment. We spent a significant amount of time ensuring we don’t lose ourselves in the technical brilliance of what we developed, but instead saw the value in the results those capabilities could bring us.
The Smartest Thing We Did to Set It Up For a Successful Launch
BP: There were three or four things, but I think the smartest thing was…our CEO [providing] clarity and intentionality around why we are here, and that is to make AI a force multiplier. I think that declaration is very, very powerful.
The second smartest thing we did was to modernize the infrastructure before introducing any further emerging technologies. That really helped us ensure we built the right secure, flexible, and scalable foundation.
Another smart thing we did was to adopt a composite mindset. I think that was very important for us, because it allowed us to flex and make sure we were blending and building capabilities that fit our purpose.
And the last thing was democratizing our capabilities, with a path-to-value mindset, by building a sandbox that mirrors all of our available capabilities but is outside our production. It helps our people become better and better and fine-tune those opportunities that tie back to very specific business outcomes.
We track metrics like call handling times, claims, automation rates, workflow cycle times, self-service usage, and speed of experimentation. It’s very outcome-based…
Metrics We’re Tracking
BP: We measure success through what I call operational efficiency and productivity, as well as customer and associate experience, and the adoption of AI tools we make available across the business.
We track metrics like call handling times, claims, automation rates, workflow cycle times, self-service usage, and speed of experimentation. It’s very outcome-based, and at the same time, we track technology metrics. But our focus is on a path to value outcome.
For example, we have seen a major improvement in our digital intake. It is up 2x, and our auto adjudication of claims is up 9x. We also reduced our claims call volume by 32 percent through simplification and automation.
One exciting thing for our call center associates is that we have reduced after-call work by 40 percent. So when they are done speaking with a customer, instead of going back to put everything down on paper to make sure it’s an accurate reflection of their discussion, they’re able to record it with the tools they have. They also receive the information in a way that they are able to properly understand it and connect it with prior history, so there has been a significant reduction in that [work.]
We are also seeing meaningful improvements in the way we develop. Not only do we use an Agile process for development, but we also have new tools that help us develop faster and more accurately, and test differently. In addition to the pure technology metrics on the platform’s efficiency, it is also very outcome-driven, tying back to the processes and activities we’ve been talking about and the products we’re putting into play in the market.
Featured image: Beyond My Ken – Own work, CC BY-SA 4.0
You must be logged in to post a comment.