Cognii Chasing “Siri for Education” at LearnLaunchX & TiE Challenge

This is a story about one entrepreneur, two accelerators, and three big ideas.

The first idea is how to make money at an education-tech startup. The second is how to do automated assessment of students’ written responses. The third is how the technology is aiming to become something like “Siri for education.”

All of the above come together in a Boston startup called Cognii. The young edtech company is set to graduate from the LearnLaunchX accelerator this week—investor demo day is Wednesday—as part of that program’s inaugural class. Cognii is also part of TiE Challenge, a year-long startup mentorship program run by TiE Boston.

Cognii is led by founder Dharmendra Kanejiya, who goes by Dee. He did his PhD work in speech recognition and natural language processing at the Indian Institute of Technology in Delhi, and worked at Boston-area startup Vlingo, assisting in the company’s technology integration with Nuance, its acquirer, in 2011-2012.

Dee Kanejiya, Cognii founderDuring his previous work on statistical language modeling, Kanejiya (pictured) had come across some research in educational assessment. As he puts it, edtech roughly breaks down into three categories: online learning platforms, content publishers, and assessment tools. He got hooked on the last one.

“Today education is in a state where the Internet was in 2000. Education hasn’t benefitted a lot yet” from technology, he says. “There’s lots of progress in instructional delivery and videos, but the second important part is assessment. Without assessing someone’s knowledge, education isn’t complete.”

Cognii is trying to create an algorithm-based way to do online assessment that can scale up to large numbers of students and courses. The goal is to help teachers and professors do their grading—not replace them—especially for online courses. The tricky part, of course, is getting a computer to evaluate written responses (not multiple-choice answers) in a way that’s reliable and accurate.

Kanejiya says he has created a half-dozen custom demos around content in areas such as biology, history, management, and music theory. A simple example of the latter would be analyzing answers to the question, “What is musical harmony?” A correct answer would include the idea that two or more musical notes are played together, and that the sound frequencies of the notes are related in a certain way.

Cognii’s software does natural language processing and compares a student’s answer to the expected answer, thereby figuring out “which concepts are present and which are absent,” Kanejiya says. The technology can also provide a score, feedback, and hints that could be used as a learning aid. The target-student range is pretty broad, for now—from fifth or sixth grade up through university levels.

You might imagine the technology isn’t perfect, especially if there are subtle distinctions in concept or unusual turns of phrase. But as long as the software is customized for different courses, Cognii’s founder says, “The technology is there. It’s mature,” he says. “We are currently trying to figure out how to go to the market.”

Cognii is looking at selling to massive open online courses (MOOCs); online publishers who want to make their content interactive; companies that make learning management software; and corporations that want to improve their employee-training programs.

So where does the Siri part come in? Kanejiya is a speech guy, and he envisions that the future of education will be “very personalized,” letting people learn at their own pace. “Mobile learning” will become a big trend, he thinks, and “that is where the speech component will come in.” As he puts it, “you can have a learning session spoken to you, it will ask you a question, and you can speak the answer.”

That kind of “intelligent tutor” is a Holy Grail of both edtech and artificial intelligence, of course. But it seems like a viable goal, especially for someone with deep experience in speech and language technologies. Indeed, Cognii’s founder says he hopes the company becomes “a hub for doing innovative research in education.”

For now, he seems to be making good use of resources around Boston. TiE Challenge has given Kanejiya valuable feedback on presenting his ideas, he says, and is facilitating introductions to angel investors.

Meanwhile, LearnLaunchX “has been tremendously valuable for me and my startup,” he says. “It gave us initial exposure to the education space, and has led to a lot of inbound interest.”

“VCs have been very receptive,” he says.

Author: Gregory T. Huang

Greg is a veteran journalist who has covered a wide range of science, technology, and business. As former editor in chief, he overaw daily news, features, and events across Xconomy's national network. Before joining Xconomy, he was a features editor at New Scientist magazine, where he edited and wrote articles on physics, technology, and neuroscience. Previously he was senior writer at Technology Review, where he reported on emerging technologies, R&D, and advances in computing, robotics, and applied physics. His writing has also appeared in Wired, Nature, and The Atlantic Monthly’s website. He was named a New York Times professional fellow in 2003. Greg is the co-author of Guanxi (Simon & Schuster, 2006), about Microsoft in China and the global competition for talent and technology. Before becoming a journalist, he did research at MIT’s Artificial Intelligence Lab. He has published 20 papers in scientific journals and conferences and spoken on innovation at Adobe, Amazon, eBay, Google, HP, Microsoft, Yahoo, and other organizations. He has a Master’s and Ph.D. in electrical engineering and computer science from MIT, and a B.S. in electrical engineering from the University of Illinois, Urbana-Champaign.