Written by Jerred Ziegler.
It’s critical that a business has enough supply to meet its customer demand. Typically demand forecasts are determined by looking at historical sales, rather than all of the potential influences such as weather, promotions and product placement. As a result, companies would often find themselves facing either excess production costs or unhappy customers. Columbus, Ohio’s Nexosis is using machine learning to reduce these errors before they happen, helping make businesses smarter with more accurate demand forecasting.
Machine learning uses computer technology to continuously analyze data over time, identifying patterns in customer buying habits by looking at many different business-driving variables. These variables include historical sales reports, the weather and even customer emotions based on past survey responses.
“Historically, the problem with machine learning has been that you need to be a data scientist or have a deep background in math to be able to do it,” said Ryan Sevey, CEO and co-founder of Nexosis. “We want to bring machine learning to everyone. We provide the platform that every company needs in an effort to do this.”
Nexosis data can be useful for companies who sell goods and those with large employee teams. For instance, if you’re a candy company and need to figure out how much product to have on store shelves, Nexosis can forecast the needed stock. If you’re a pizza shop and need to determine how many delivery drivers to have on hand at a given time, Nexosis’ technology can help predict the human resource needs before they happen.
Sevey and his co-founder Jason Montgomery developed the idea for Nexosis based on research they did in 2013 regarding the prevalence of cheating in video games. Using machine learning, the two were able to develop a system that could analyze a player’s data to determine if they were cheating in the game. They presented their findings at a conference in 2014, where they were first introduced to Rev1 Ventures, a central Ohio partner of Ohio Third Frontier.
“We had no intention of becoming a startup business until after that conference,” said Sevey. “Once we began talking with Rev1, we really started thinking about how this idea could be useful for any company. Anyone with a challenge such as demand forecasting or personnel planning can use Nexosis.”
Rev1 awarded Nexosis a $100,000 concept investment, along with additional funding from an independent investor. These early investments allowed Sevey and Montgomery to evolve their research into a product that could be sold to any business as an easy solution for machine learning.
Nexosis rented office space at Rev1 in summer 2015 before moving to its own independent space in Westerville, Ohio, in the fall. The company currently has seven full-time and six part-time employees, with the goal of having a staff of 24 by the end of 2016.
Both Sevey and Montgomery are Ohio natives and agree that the low cost of living and the large talent pool make Ohio ideal for starting a new business.
“It’s exciting to see the talent we’ve attracted so far. We’re located near some of the top schools in the country for machine learning graduates, like The Ohio State University, Carnegie Mellon and the University of Michigan,” said Montgomery.
“We always tell any venture capital firms looking to invest in Nexosis that we are not interested in moving out of Ohio,” said Sevey. “It’s a great time to do business here, and the companies in Ohio are really starting to show the rest of the country what we’re capable of. I wouldn’t be surprised to see more venture capital firms looking at Ohio for their next investment.”