Where’s the Innovation?
We’ve all seen disruptive innovations in action: a groundbreaking product comes in, shakes up the industry, puts leading firms out of business and creates an entirely new market. Think Uber, the cloud, and entertainment-on-demand—all of these things have transformed long-held ideas and turned their respective industries upside down.
But what about lenders and insurance companies? Have we seen anything that has revolutionized the way we go about getting approved for a mortgage, for example? Not really. It’s still a grueling, multi-day process.
Despite the fact that lenders and insurance companies have made significant investments in technology, the industry hasn’t yet experienced wholesale technology disruption. Banks operate the way they’ve always operated. And while you might argue that there are “fintech” disruptors such as Lending Club, Bitcoin, Prosper, and others, these are more on the borders, as servicers who trade in information rather than loans, policies and currencies.
The Reason Lurks Beneath the Surface
The absence of disruptive innovations for lenders and insurance companies is not because the technology doesn’t exist yet. The technology is out there, just waiting to be harnessed. Rather, the reason for this lurks beneath the surface of these companies’ operating models. Typically, their processes are complex, and the level of precision with which they define these processes isn’t precise enough.
But why do lenders and insurance companies have all of this complexity? A lot of it centers around the vast variability of data that the industry operates on. Essentially, to apply precision and reduce complexity, we need to go from unmanageable differences in the way we describe borrower, property, insured and circumstance data; down to a cookie-cutter sameness that everyone can understand and work with.
And yet, the process to funnel data down from the vague and varied to the same is weighed down by all kinds of ambiguity, which persists far further into the process than it necessarily has to; and the number one cause of that persistent ambiguity is unstructured data. Once the problem of unstructured data is addressed, it becomes feasible and worthwhile to make the effort to precisely describe processes, logic and the data necessary to inform them.
The Problem of Unstructured Data
In the past, the only way to get unstructured data into a structured form was for someone to key it in. But the amount of data that needs to be keyed in is much larger than it appears to be on the surface. Typically, the decision-making process involves more than just facts and figures. For example, when decisioning a loan, it’s not enough to know that a property was appraised at $200,000. We also need to know who performed the appraisal, when it was done, and other important “metadata” about the appraisal. The metadata allows us to make a wider judgement about the number $200,000.
However, the cost and effort involved in determining what the metadata should be, and then keying this mountain of data and metadata in, is so heavy, that it hasn’t make sense to invest in it. Until now, the only practical way of dealing with unstructured data has been to leave it unstructured, and have human analysts use their sometimes “tribal” knowledge to review the relevant data, and ignore the irrelevant data. Ultimately, the lack of precision in the industry is in part because the ROI on making things precise has just not been there.
Machine Learning Can Bring Structure to Unstructured Data
If you’re a company operating in the financial domain, and you’re buried beneath a mountain of unstructured data, how do you unleash technology in a successful way? First, you need to bring structure to unstructured data, and precision to the ambiguous processes in your operating model. Then, you need to apply the right technology to automate the solution in the most cost- and time-efficient way.
Machine learning, or AI, is the best way to liberate data from the unstructured form that human communication and commerce is based on, to a structured form that machines can understand. AI has the ability to use algorithms to apply precision to unstructured data, and to increase the accuracy of those algorithms by learning from human feedback. Once the data is in structured form, the decision-making process becomes a lot easier. Many industry participants and tech suppliers would then be able to use the data for business rules, calculations and other job functions that currently depend on analysis performed by people.
A Digital Workforce to Disrupt the Lending and Insurance Industry
Here, at HeavyWater, AI is our bridge from the highly variable, to the distillation of common structures and processes. Using AI, we introduce structure and precision, first by targeting unstructured data and distilling it to a structured form that has meaning in a specific financial sense. Next, we focus on ambiguous processes, and describe how these processes work using very detailed blueprints. Once we’ve transformed the ambiguous and unstructured to the simple and meaningful, we deliver it to other systems that perform analysis functions currently limited to people.
Essentially, we offer an inexpensive digital workforce of assistants who are trained to perform processes in a disciplined way. Our virtual assistants quickly and competently rationalize a mountain of unstructured data into something that makes sense to everyone, which creates a significant value proposition for any company interested in maximizing their profits.
So, if you’re a lender, an insurance company, or a similar entity looking to innovate, drop us a line. We offer solutions that will turn the lending and insurance industries upside down, and give you the competitive advantage that only disruptive innovation can offer.