By Scott Meacham
Copyright © 2017, The Oklahoma Publishing Co.
Did you ever have an idea that just wouldn’t go away?
That’s the story behind iRecommend Software, a Tulsa-based artificial intelligence and machine learning startup. Founder and CEO John Morad started writing software when he was 8 years old, and has been writing software ever since.
While he was bootstrapping and building a very successful software consulting business in Tulsa, he had the vision for the software that is now iRecommend.
It started with personal assistant software that Morad coded to collect information from the public domain to help individuals make decisions. By 2010, Morad had progressed to a recommendation engine that reached about 60 percent accuracy in its recommendations.
He kept at it, developing an artificial intelligence agent that creates models dynamically, achieving better than 80 percent accuracy.
“As a human I am imperfect, and as a software developer, I cannot define and build something that is perfect,” Morad said. “But if I can build an artificial intelligence agent — basically a robot without the physical appearance — that AI agent can build precisely the software I need.”
Early iRecommend customers are in retail, employment recruiting, and real estate. Sophia (the AI agent kept her original name) provides algorithms that put the right product, job, or house in front of the right person at the right time. That’s a plus for consumers, job seekers and employers, and families looking for a place to live, with iRecommend supplying as many as 10 to 20 million matches per month.
These days, everything is smart — our devices, our cars, our home security systems. We have come to expect that whatever we need will be at our fingertips instantly — whether it’s a show on Netflix, a book on Amazon, or that special gift we are looking for online. However, it requires complex analytics for online retailers to quickly and accurately determine what we want.
Search engines provide generic results. There are so many options online, from products, to entertainment, to jobs, that neither buyers nor sellers can assimilate it all. Imagine if instead of generic results, you received information that was specifically (and reliably) tailored just to you — information that matched your needs or desires better than 75 percent of the time.
“There is so much public data available today that we should be able to recommend things at the right time and price that people need in every area of their lives,” said Morad. “Most of data analytics gives business and consumers the map to find the gold. We want to skip the map and just give the gold.”
Scott Meacham is president and CEO of i2E Inc., a nonprofit corporation that mentors many of the state’s technology-based startup companies. i2E receives state support from the Oklahoma Center for the Advancement of Science and Technology. Contact Meacham at [email protected].
