When you just have to have a blue coat with pink lining …
Startup wants to make ‘No results found’ a thing of the past for online shoppers
WATERLOO REGION — A University of Waterloo grad is going public with technology that uses artificial intelligence to improve search results for online shoppers.
G Wu is taking his latest startup, Adeptmind Inc., out of stealth mode and is building a North America-wide network of PhD candidates who are researching artificial intelligence. He will use some of the $5.5 million in venture funding he has raised to pay for 20 scholarships, each worth US$30,000.
Wu said Adeptmind has developed software that is inserted into a retailer’s back-end system for online sales, ensuring its products are displayed in response to queries from shoppers.
“The reason we are out of stealth is we did some prototypes and this 100 per cent actually works,” said Wu.
Online shoppers are routinely frustrated by “no results found” and end up going to Amazon or Google, he said. But with Adeptmind’s technology, small and medium-sized businesses can compete for online sales with the tech giants.
“Out-of-the-box we get better-than-Amazon accuracy,” said Wu. “And then we have various techniques to boost accuracy further after launch.”
Machine learning kicks in after Adeptmind’s technology is deployed, quickly bringing the results to 100 per cent accuracy.
“The advent of current research in deep learning, in reinforcement learning, in incremental learning, in active learning make this possible,” said Wu.
With Adeptmind software, shoppers can search for winter parkas with fur-lined hoods and midsection belts that are red, and be assured of finding links.
Twice a year in Las Vegas, the retail fashion industry gathers for a huge convention called MAGIC. During the last convention in August, 34 users tested Adeptmind’s software, trying to stump it with very specific and obscure search requests. “And nobody broke our system,” said Wu. The current search technology for online shoppers is so bad the same level of performance would not be tolerated by consumers in other sectors of the economy, said Wu.
Between 1992 and 1995, hardly anyone searched the internet. Internet search did not really take off until Google PageRank was launched in 1997. By 2000, Google was on its way to dominance.
“We are hoping to demonstrate to the world that e-commerce search at 100 per cent accuracy, while it does not sound too sexy, it actually is really sexy,” said Wu.
“It is as sexy as when Google did PageRank in 1997 and searching the internet went from ‘nice to have,’ to ‘absolutely needed,’” he said.
Wu, Adeptmind’s chief executive officer, is well known in Waterloo Region’s startup community, but for the past 18 months he has maintained a low profile.
He was a founder, chief architect and vice-president of business operations at Maluuba. He oversaw the Waterloo startup’s research lab in Montreal, which was focused on natural language processing, and left about seven months before Microsoft acquired Maluuba in January of this year.
Wu founded Adeptmind with chief technology officer Jing He in September 2016. The startup’s corporate office is in the David Johnston Research & Technology Park in Waterloo. Its operations office is in Toronto, where 20 software developers work. Adeptmind raised $5.5 million in funding in March.
The startup is working to make sure its software quickly integrates with e-commerce platforms such as Shopify, Magenta Boutique, Oracle, Salesforce and BigCommerce.
“As of this moment we will be building out those plug-ins, starting with Shopify,” said Wu.
Wu said his experience at Maluuba was very helpful.
“When you start again you know how to raise money, you know how to hire people,” he said.
“When we were kids back in the day we didn’t know how to run a business. Nowadays it’s the only thing I know how to do.”