Ex-Top Hat founder’s edtech startup sells peer-grading and AI software to university professors.
When every university student can have an AI chatbot whip up an essay draft for them, how can professors ensure students are still learning critical thinking?
The new capabilities are pushing educators to explore different means of assessment—with and without AI—cracking the classroom door open for edtech companies to step in and suggest new ways of using technology to learn.
“Kritik is moving education from ‘did the student get it right?’ to ‘is the student getting better at thinking?’”
One such startup is Toronto’s Kritik. Founded in 2019 by Top Hat founder and former executive Mohsen Shahini, education researcher and repeat founder Carine Marette, and tech executive Mark Deepwell, Kritik’s pitch to university educators is that building trust with students and allowing them to use AI is the best bet.
At digital education platform Top Hat, Shahini noticed a shift happening in student assessments, with less content memorization and more discussion posts. Shahini wanted to build a peer-to-peer assessment tool where students could collaboratively grade each other’s work. So, he decided to create a standalone product instead.
“I started feeling that there was a need for assessment of the students at a deeper level, at the critical thinking level,” Shahini said in an interview with BetaKit.
Kritik got its start just months before the COVID-19 pandemic shut down schools. Assessments moving online gave the fledgling startup a boost. The company kept building up its product, but then came ChatGPT. Suddenly, some professors hesitated to assign the writing assignments that Kritik’s software facilitated, because they couldn’t tell if the student had used AI.
Today, AI models have only gotten better at generating passable, even convincing, writing. AI detection tools, “humanizers,” and other products have cropped up to beat back the wave of AI-generated student work, which Shahini likened to an “arms race.”
“It’s kind of a lost battle,” he said. “We decided that our approach is different…less about policing the students and more about transparency and building a dialogue.”
Kritik’s latest product, VisibleAI, doesn’t work as an AI detection tool. For Shahini and his co-founders, that’s not the point: they want to give students and professors flexibility and visibility into how much AI can be used and when. Depending on the professor, Kritik’s product is built with the expectation that students are supposed to use AI, and how effectively they do so becomes part of the evaluation. The student works on assignments within the VisibleAI window, with full access to different LLMs. It tracks what parts the student types directly, what prompts the student is using, and how much of it is pulled from the AI chatbot.
RELATED: Top Hat CEO Mike Silagadze on EdTech’s crazy school year
Edtech is a $164-billion USD industry. But some studies have found that increased dependence on AI is associated with lower levels of critical thinking. In 2025, MIT researchers tested people using AI to write an essay and found that those allowed to use ChatGPT progressively got lazier and exhibited lower levels of brain activity. Delegating mental tasks to AI has been termed “cognitive offloading,” and in one study, negatively impacted the long-term buildup of coding skills.
For Marette, who is completing a doctorate in education, that’s a problem Kritik is trying to mitigate. “Some people are afraid of AI, some people embrace AI, but there is an overusage of AI,” she said. They’ve designed VisibleAI to discourage over-reliance on the technology and instead make the process of writing more transparent for instructors.
“When the AI is constrained, the students are using more of their creativity and technical skills to provide higher quality [work], as opposed to cognitive outsourcing,” Marette claimed.
Kritik has also expanded its original peer-grading product, Kritik360. Rather than autonomous agents and AI-generated feedback, Kritik360 allows students to grade each other’s work and claims that humans are kept in the loop. It also allows students to review and critique each other’s use of AI, Shahini said. However, it also offers an AI course creator based on uploaded course syllabi.
“Kritik is moving education from ‘did the student get it right?’ to ‘is the student getting better at thinking?’” Marette said.
Kritik doesn’t want its AI software to replace teachers because that’s who its end customers are. It sells its product at the course level, often to individual instructors, and says that Kritik has been used in courses at more than 250 institutions—including the University of Waterloo and the University of Alberta.
Shahini claimed that Kritik is at a cash flow break-even point, and has stopped actively pursuing VC money after raising $3 million in 2022. The company has over a dozen employees and hopes to reach 25 people by the end of the year.
Feature image courtesy Unsplash. Photo by Zoshua Colah.
