Integrating AI learning into insurance processes is happening right now.
And right now, too.
And also, right now.
We all know data feeds AI. The more you feed your models, the better they get at discovering patterns and anomalies, predicting customer behavior, assessing risks and value and so forth. These capabilities can be quite mind-blowing, really. So why are so few companies doing it? In many cases the problem isn’t a shortage of data. The problem is instead adding new to old: integrating new data, learnings, and models into old, legacy systems.
Even companies with dedicated and skilled data science teams find that turning a model from an academic exercise into something that affects results is a formidable challenge. Why? Because companies that have a vast amount of subject matter expertise also frequently have IT infrastructure that drags them down. Legacy systems can severely limit access to both data and system connections, leading to a situation where it’s difficult to develop models using the latest open-source AI & ML technology, and even more difficult to integrate models into any business process.
It’s a conundrum. But not an unsolvable one. In fact, we’ve solved it for our insurance partners.
Partners working with Slice have the ability to take models developed by their own in-house data science teams and deploy them quickly and securely on the Slice insurance cloud services (ICS) platform. Our ICS platform, with its microservices architecture, makes it easy to plug in models at any point in the front or back-end, and especially at any point in the customer journey.
In short, combining the expertise of a partner’s data science team with the flexibility and security of our ICS platform means models can drive action and get there fast.
Let’s look at our partner, Duuo (a digital offshoot of The Co-operators). Their internal team tackled a range of problems related to assessing the risks associated with homesharing (short-term rental) activities. Once they had finalized the model, it was securely sent to the ICS platform to be integrated, tested, and deployed to production. It is running live and in-flow right now. Our platform provides the model results, computation times, and observed testing data back to our partners to enable them to connect the full feedback loop.
So how does that solve the problem of integration with existing IT systems? Often there are relatively few ‘integration points’ where the results of a model can be used. Sometimes there’s a CRM (customer relationship management) or marketing automation system, but generally speaking there is a shortage of ‘plugs’ for machine learning models to hook into.
By way of contrast, the ICS platform is all plugs. Models are used at every point in the process:
- improving customer onboarding flow,
- analyzing conversion,
- evaluating risk, and more.
Our microservices architecture allows us to integrate models in-flow thereby making it easy to use a series of smaller, focused models to solve problems. So, rather than putting all your eggs in one basket to solve one huge problem and dealing with the complexities inherent in black-box modeling, you can instead use many smaller baskets to accomplish the same goal much more efficiently.
Using a series of models to solve problems, and making it simple to deploy, update and monitor the results is the magic that lets us integrate AI and machine learning securely and quickly – even for our partners with legacy IT systems.
When it comes to adapting insurance processes to today’s on-demand consumers, it’s not so much “out with the old and in with the new”, which should be comforting, since getting rid of the old is a long and arduous journey. All it takes to harness your data, feed your AI machine, and respond to your customers’ needs, is a partner who can help you integrate what you already know. Learn more about Slice Mind here.
