Case study: Relativity6’s industry classification tool

Case study: Relativity6’s industry classification tool

Ringvald is counting on generating some major traction, in part, by solving that frequent small commercial underwriting data snafu.

Venture capitalists are intrigued. The company recently announced that it had raised a $5.25 million seed round that included Fin Capital, State Auto Lab Funds and Vectr Ventures among its investors. Plans call, in part, for using the money to boost sales efforts within commercial insurance underwriting and expand into new financial services markets. Including the seed round, Relativity6 has raised approximately $8 million to date.

Ringvald founded the company with some former classmates from the Massachusetts Institute of Technology in late 2016. About 42 people work for the Boston-based company at present, with employees also based in Los Angeles and Mexico. Relativity6 has 45 customers and counting, Ringvald said.

The tech

In layman’s terms, Relativity6 has developed an underwriting tool that helps predict what a company’s primary business is. In real time, Relativity6, working with a customer via its API, takes the name and address of a business and then detects what a business does. Then, it predicts the top two mostly likely six-digit North American Industry Classification System (NAICS) codes in a process that takes about a second, Ringvald said. It also comes with confidence scoring, so customers know exactly how accurate the predictions are. In addition, the tool gives users evidence as to where the predictions came from.

Relativity6’s primary customer target is a commercial underwriting group at an insurance carrier that focuses on small businesses because “small and midsized businesses are very difficult to classify.” The reason: small businesses rise and fall rapidly so traditional databases struggle to keep up, Ringvald explained.

“Usually you have to do it by hand,” he said. “That’s actually our strength, being able to classify mainstream smaller businesses with high accuracy.”

Carpenter versus roofer

Commercial insurers gain from a company like Relativity6 because it is as precise as possible when classifying an insured and the risks involved while crafting coverage, Ringvald said.

His example of how that plays out involves comparing a carpenter to a roofer.

“They’re pretty similar, but from a risk perspective they’re incredibly different,” Ringvald explained. “You have to know, are they a carpenter, or are they actually jumping up on a roof where they have the potential to fall off? You’d have to price that risk quite differently, and that’s something that gets missed a lot.”

Beyond an API, the company’s cloud-based system looks up businesses in real time, without the need for third-party databases. AI fuels the tool, specifically through natural language algorithms.

Machine learning plays a role as well.

“The process starts by feeding our models very clean data … we’ve done that millions of times to teach our machine good examples of these types of businesses,” Ringvald said. “The second piece is on the natural language. Once we do the real time search, and then we feed it back with better data as we go, it constantly gets more accurate.”


Commercial underwriters from carriers are the primary users, as well as comparative raters – entities who would use a small business insurance platform where brokers or end users generate quotes to get insurance (larger brokers are also customers).

Connecting a carrier to Relativity6’s API is relatively easy, Ringvald said.

He explained that one API gets integrated right into the new work submission workflow of these platforms.

“Brokers go in to try to generate a quote, and what happens is they’ll start typing in the name and address of the insured, which is all our API needs,” Ringvald said. “In about one second, our API will do what it does, and start predicting the most likely six-digit NAICS class codes of that business.”

Meanwhile, as the user finishes typing in the name and address, the predictions come through and then the broker will confirm the insured’s proper industry.

“And now you don’t have to worry about the industry classification [as] we have that correct,” Ringvald said.

The idea is to weld simplicity with clients, hence Relativity6’s single API. Ringvald said integration typically takes place in an hour or less.

“Insurance has other issues that can make it not as fast, but it is a very simple API to integrate into a workflow,” Ringvald said.

More codes

Relativity6 sees many other opportunities for clarifying codes in the NCCI and ISO spaces, among others. There will also be further work on flagging keywords to look for, to help insurers avoid underwriting trouble spots.

“There isn’t really a good solution for identifying the industry that a customer base is in with a high degree of accuracy,” Ringvald said. “We can continue to tackle that segmentation problem that exists … It’s a global issue.”