Company’s tool automatically detects and redacts important phrases in documents.
Read in 1 minute
Two years ago, Ron Glozman found himself in a typical student situation. He was studying computer science at the University of Waterloo and business administration at Laurier University – the two schools have a joint program – when he found himself up against the exam wall.
He’d put off studying and had 1,500 textbook pages to go over in just 48 hours. Knowing the task was impossible, he instead tried to create a shortcut. He quickly whipped up an algorithm that could scan and digest the pages, then deliver him a summary. His software tool would effectively give him the Coles Notes version of the material.
Glozman didn’t just pass his exams, his friends started asking about the algorithm. He decided to make it available as a mobile app, Knote, and within two weeks he had 2,000 downloads in 33 countries. He knew he was onto something.
Students, however, weren’t a good target market.
“They don’t like to pay for technology,” Glozman says. “They’d rather spend their money on social outings, like going to the bar.”
Knote the company was born in 2016, with the new focus being on lawyers, an increasingly popular target for AI developers. With so much of their jobs being document digestion, it’s a field that’s particularly well suited for help via automation.
Glozman rejigged his algorithm to analyze and summarize legal documents. He then added another function – automatic redaction – that is now accounting for most of Knote’s business. Lawyers and especially government employees are finding the tool, which can automatically strip out sensitive information from documents, to be useful:
The federal Canadian government, for example, pays an average of $5.26 to redact a single page, Glozman says, whereas Knote’s AI brings the cost down to $1 a page.
Redaction, he adds, is an expensive activity because it isn’t a job that typically belongs to one specific individual. The government, for example, typically pays economists overtime to redact studies they have written that are to be released publicly through access-to-information requests.
“He has to go back with a marker and cross out all this information,” Glozman says. “We want to free up his time so he never has to work overtime again.”
Knote, based in Toronto, has only two employees, but is looking to add another five. The company won money last spring from the University of Waterloo’s Velocity Fund and has received a grant from the Ontario Centres of Excellence. Glozman says one of the largest law firms in Canada is a customer, but a non-disclosure agreement prohibits him from naming the company.
He believes AI can help people work at more valuable and fulfilling jobs, rather than at dreary tasks like document analysis and redaction.
“Our model is work smart, not hard,” he says. “We’re trying to move people up the pyramid. It’s going to reduce the number of people at the bottom and move them up to jobs that can’t be automated.”