Video: Continuous Intelligence in the Insurance Industry

Adrian Bowles: Here with Cathy Reese at IBM, you’re a Partner in Advanced Analytics in the Services Organization. We’ve talked offline about a number of things that are going on at IBM with your clients adopting advanced analytics. Maybe we could talk for a minute about the business value that they see that’s driving them to continuous intelligence solutions.

Cathy Reese: All the industries are looking at this continuous intelligence. I think about one company we worked with, Groupama, they’re an insurance company in Italy. And they knew even a couple of years ago that telematics was the way to be able to serve their customers even better in the future.

So they invested heavily in that, and they brought us on board a couple years ago to help build them the platform. How are they going to be able to stream in this telematics data and then quickly match it up to claims data to be able to process claims faster? Or even to be able to stream the data in and understand if someone’s been in an accident or understand if there’s some fraud or the car has been stolen. Being able to ingest the important data and then use it throughout the enterprise as a way to keep continuous intelligence kind of forming to help serve the clients better.

AB: So the cost is more reflective of their actual individual case rather than some broad class that they’re part of?

CR: If they could serve the customer at a lower cost than they can lower the price that they charge out to their clients and it’s a win-win for everyone.

So the trouble with telematics data was before that it kind of sat in a silo by itself. There were some insurance companies that I’ve worked with where it was more of a marketing thing, and so they never even used it. They never even shared that data with the actuaries. So what we really had to do was come up with a platform like IBM Cloud Pak for Data to be able to bring together the historical data, the streaming and telematics data, the customer service data, the policy data, so people can actually make real-time decisions off of that information.

I think technology’s evolving so fast and our clients are really worried about that. They’re not really sure where to invest. We’re helping them build the right flexible architectures to be able to meet the use cases of today, but also think about how they’re going to meet the use cases of tomorrow. I think having IBM Cloud Pak for Data, along with the data platform and all the tools that are offered and available as part of that package solution, is one way that we’re really helping our clients solve for that problem. It takes a little bit of the anxiety out of that decision making as they’re going through this huge investment and this huge transformation.

AB: Right. So even if you’re only using a few of the features today, you can be comfortable that you can build on that going forward.

CR: Absolutely.

AB: Okay. And in Cloud Pak, it’s obviously something that’s coming up a lot as I talk to people at IBM. Everybody seems excited about it. Maybe just talk a little bit, since you’re in the services side, as you’re developing systems and the frameworks for your clients, how much of that are you able to reuse and leverage in other engagements?

CR: Sure. I’m working with a lot of clients who have kind of been stuck in this proof-of-concept world. They run a lot of experiments, they find answers to questions, but they’re not actually getting those models all the way up into production and really industrialized throughout the enterprise. So we’re working with them to help build some common frameworks of how to build analytics and build assets and models so they’re repeatable or usable and they’re actionable, they actually drive outcome, and we’re not continually rebuilding things over and over and over again. Or we’re not hard coding things in our laptops and not sharing them with the other data scientists within the organization.

So we’ve come up with a framework to help them build that. We use the IBM Cloud Pak for Data with Watson Studio to be able to help get those models shared appropriately. AutoAI to be able to do some of the data science for us, and even OpenScale to be able to watch for bias and to retrain the models and to be able to understand what’s happening to them once they’re in production.

AB: Okay. You bring a bias, that’s obviously a really hot area right now. Is that something that clients are asking about or are you sort of proactively saying, “Here’s something to be aware of?” I mean, I think if you look at insurance as one of the examples, you can think off the top of your head of a lot of ways that the data could be biased.

CR: Absolutely, I mean ethical AI is something that we’re really working very closely with our clients on. They want to make sure they’re monitoring the models for bias because we inherently all have it. So to be able to use software to be able to manage that I think is important. And even just to manage the models for drift or things that kind of can start them going off awry. So using some tools to help clients do that is something that they’re very interested in talking to us about.

AB: The idea of drift is fascinating to me, and when you talk about continuous intelligence, drift is something that you have to be looking for.

CR: Yeah, absolutely. I mean, people really need trust. Trust is an essential part of where we’re going with AI. You’ve got to be able to trust in the data. The customers have to be able to trust in you that you’re going to manage their data the right way. So trust is a core part of doing that.

AB: Great. Well, thank you. This has been good.

CR: Thanks.