This startup takes the work out of data analytics

Andreas Rekdal
August 3, 2016

Doing data analytics is a lot like striking up a conversation: you’ve got to ask the right questions if you want to learn anything interesting.

But in many organizations, the employees best equipped to ask the right questions — the ones who are close to day-to-day operations — lack the ability to investigate their hunches. Unless they can convince higher-ups that researching a hypothesis about consumer behavior or business operations is a worthwhile expenditure of limited resources, their suspicions may never be confirmed or set aside.

Semantify, a Chicago-based analytics company, wants to make investigating business hunches easier. Their plan for doing so is to let non-technical users do their own analytics by typing in plain English questions like “How many customers who call to complain about late payment fees end up closing their accounts?” into a simple search bar. The platform automatically generates a report in response, providing charts and graphs to illustrate trends and distributions if applicable.

“The way this has to be done now is that the businessperson will have a question, and there are ten different sources from which the answer has to be pulled by data experts,” said CEO Ashoke Dutt. “That would take days, and you’re waiting for that answer to ask the next question.”

A former bank executive with experience from Citigroup, Morgan Stanley and Discover, Dutt first got involved with Semantify as an angel investor in 2008. On assignment from a number of large banks, the company started building a platform that would make it easier for bank executives to perform ad-hoc analysis of their companies’ extensive financial data.

Moving beyond finance

Thanks to its early bank partnerships, Semantify had access to huge quantities of financial data while building the machine-learning platform. Using this data, the team trained its software to navigate complex data structures and interpret jargon-ridden user queries.

But once the platform got a good handle on financial data, Semantify realized that much of the knowledge embedded in the platform was equally applicable to other industries like healthcare and insurance. Consequently, the company broadened the scope of its ambition, expanding into new verticals and making the platform available across departments and levels at any given organization.

“Whether you’re a call center representative or a CEO, you will have an application where the knowledge model has captured all the business knowledge of that domain,” said Dutt.

Estimating about 80 percent of the platform’s knowledge base to be reusable across industries, Dutt said Semantify can be adapted to understand the terminology of a new industry in a matter of weeks. Introducing the platform to a competitor in the same industry can be done in mere days. And after the initial adaption is done, customers can operate the platform on their own.

“What we wanted to do was to disrupt the business intelligence industry’s model of consultant-based revenue,” Dutt said. “We want this to be fast deployment and self-service. After they buy the product, the customer need not even call us.”

Building that ground game

Though its existing team is distributed across several continents, Semantify is establishing its new headquarters in Chicago, where Dutt has been located for years. The company’s board of directors includes Chicago notables Jai Shekhawat, the founder of Fieldglass, which was acquired by SAP for $1.1 billion in 2014, and KDWC Ventures partner Chris Capps, who also serves on the boards of companies like HighGround and Rippleshot.

With the codebase of the application largely complete, Dutt said the company’s primary focus moving forward will be on sales, generating knowledge models for new industries and adapting the software to run in the cloud.

Though he has no doubt that other companies will be making moves to make analytics tools more user-friendly in coming years, Dutt believes competitors will have a hard time catching up, even if they figure out how to build a competing platform. Just like human learning, machine learning takes time.

“You can’t create it overnight,” he said. “It’s not enough to create the code. You have to make it wise by ingesting all this stuff, and that’s been done over five years.”

Images via Shutterstock and Semantify.

What's your company's story? Let us know with a tip or a tweet @BuiltInChicago.

Chicago startup guides

Best Companies to Work for in Chicago
Coolest Offices in Chicago Tech
Best Perks at Chicago Tech Companies
Women in Chicago Tech