In December, Novarica published a report about intelligent text ingestion, covering the emerging and flourishing space of converting documents and PDFs into digital data. Despite a decade or more of digital modernization efforts, much of the insurance industry is still driven by document-based processes. Whether these documents are physical (via mail or fax) or virtual (via PDFs or scanned images) the handling of them has been manual, requiring human beings to look at incoming documents and retype their fields into a system in order to convert them to structured digital data.
Insurers, vendors, and industry standards groups have attempted to solve this cumbersome problem by looking for ways to get rid of documents altogether. If agents, brokers, insurers, and reinsurers can directly integrate with each other for the transfer of digital data, then documents would no longer be necessary. Insurers have built agent portals where agents can enter submissions directly. Standards bodies have created object models intended to generate universal language for easing the transfer of data between groups. Vendors have created exchanges to encourage a single point of integration for the industry. And many different parties have attempted point-to-point integrations, a difficult and unending task.
Yet after all of those efforts, a substantial amount of business in nearly all lines, and a majority in some lines, is still transacted via documents requiring manual processing.
It turns out that an emerging area of artificial intelligence (AI) might end up being the silver bullet that allows the industry to bypass a complicated integration solution while still enabling structured data to flow smoothly between all parties. There is now a broad set of options that use a combination of optical character recognition, natural language processing, and machine learning to turn documents into digital structured data with high confidence in the results. This allows incoming documents to immediately go into an insurer’s systems and processes without extensive manual intervention.
This is a huge opportunity for the insurance industry, with impact to operational costs, cycle times, data accessibility, and customer satisfaction. It’s no wonder so many vendors, both startups and incumbents, are rolling out solutions for insurers, some of whom are profiled in Novarica’s recent report. For insurers investing in this area of emerging technology (and any insurer with a “document problem” should at least investigate it), it’s worth considering what the technology and vendor space will look like in the future.
The next few years will see ongoing technology investment in this space, which will be led by the horizontal technology giants like Microsoft, Google, and AWS. Not only do these companies have the money to iterate and improve this tech, but they also have a massive amount of data to train and evolve their systems. In the AI space, access to data is what wins the day. It’s unlikely that any startup or even a large-but-not-quite-as-large vendor will be able to compete on the underlying technology. Sometime soon, the ability to take a PDF or document scan and convert it to digital data will become a commodity. The best way to do this with high confidence will be through one of those horizontal platforms.
Does this mean the growing space of intelligent text ingestion companies will go away and insurers will simply run their documents through Azure, Google.AI, or AWS? No, because converting a document to data is only part of the problem. The opportunity that remains is taking that raw digital data, understanding its insurance context, and moving it into and through the insurance business process. The vendors who win this space will be those who do more than convert documents—they will be the ones who best add a layer of insurance industry intelligence on top of the technology.
We will see a similar evolution in any other area of AI that is driven by machine learning interpretation of massive amounts of data. Systems that evaluate images of car accidents or roof damage, systems that root out fraud, systems that identify sentiment in call center interactions… the underlying tech will be superseded by the big players with more data and more computing power. What insurers will need are vendors who can apply the platform technology to insurance industry use cases.
Add new comment