Knewton Personalizes Learning with the Power of AI

Knewton uses AI to tailor educational content to the needs of individual students, allowing for a more personalized learning experience that helps students achieve mastery of learning outcomes.

What if each and every student was able to progress through their learning materials at their own pace and in their own style? Knewton believes that the answer is simple: students will learn more, and the students that will benefit the most are the ones that are struggling the most [1]. Knewton is an AI-powered platform that leverages curated education content to tailor personalized lesson plans to students. Knewton has created a lot of buzz in the education space. It raised more than $180M in investment capital, and it has made more than 15 billion personalized recommendations to students since its founding in 2008 [2].


Knewton’s technology has a track record of success with students. In a study of more than 10k students, Knewton was able to show that demonstrating mastery of subjects by completing its program’s assignments was directly tied to better performance in courses overall. This benefit was even larger for students who were struggling, persisted over time to future assessments, and increased the pace of learning.

Source: 2017 Student Data Insights, Knewton [3]

AI is at the core of Knewton’s value proposition, as its ability to personalize content to students is dependent on its AI algorithm. Knewton trained its product with over 25,000 beta testers to identify knowledge gaps in users and figure out what content worked best at remediating those knowledge gaps [4]. Knewton has curated its content primarily from open education resources (OER) to lower costs for students, but also allows teachers to upload their own content into the platform and tag it to Knewton’s taxonomy so it too can be leveraged by the algorithm. In a traditional education environment, a teacher is limited in their ability to personalize content to individuals as they cater to dozens of students at a time. In contrast, Knewton attempts to create a faculty-to-student ratio of 1:1 for every single student with its degree of personalization. This “robot tutor in the sky” has been used by more than 300 colleges and universities in their effort to create value in the form of student success, especially for those students most in need of help.


Because 90% of Knewton’s educational content is licensed through open educational resources, it is able to keep its content costs remarkably low [4]. Students are able to purchase Alta, its AI-backed higher ed platform, for just $40 versus the typical cost of a college textbook. OER content is by definition free, so that leaves almost all of that revenue as value captured by Knewton.


Despite its technological achievements, Knewton is not without its struggles. Knewton was recently acquired by a traditional publisher, Wiley, for an estimated $10-25M, coming down from a high of investor sentiment and industry excitement that led to $180M in funding over its first several years of existence. The reason behind the bubble bursting is not obvious, but it has several potential causes. First, Knewton’s pivot to Alta software came after many years of working as an AI-platform for publishers, not as a standalone courseware product. One of Knewton’s earliest partners was Pearson, another traditional education publisher, that decided in 2017 it would rather take its adaptive learning platform technologies in house [5]. In this case, Knewton seems to be a victim of co-innovation risk where the partner it picked had too large an opportunity in competition with Knewton than to continue to act as an investor and partner. Additionally, Knewton’s technology, and adaptive personalized learning itself, has been called into question broadly. While some industries may have an appetite for “mystique,” education is not one of them, and some leaders in the industry have called Knewton’s blackbox approach a type of “snake oil” [2]. While the results do seem to speak for themselves, Knewton does seem to have undercut its own success by not making its technology accessible to its potential users.


With its recent acquisition by Wiley, Knewton has the opportunity to re-establish itself in the education sphere and tap back into the potential successes of its early start up days. First, Knewton with Wiley has a much stronger position to fight off integrated players like Pearson attempting to dominate the adaptive learning market on their own. Wiley’s core business is educational content, and if Knewton can tap into the vast educational resources and learning science that Wiley has, it has a much better opportunity to compete. In particular, Knewton can leverage Wiley’s datasets (e.g., student learning assessments) to train up its algorithm much more quickly and cheaply than its previous beta user-led approach. Second, Wiley is a credible voice in the education market as a long-tenured, traditional publisher and has deep contacts and relationships with the customer base that Knewton needs to sell to. By leveraging Wiley’s brand and sales force, Knewton can overcome the “snake oil” hesitancy that may have been fighting its efforts to get traction as a standalone company.









Gong: Conversation Intelligence Platform Sales Leaders Rave About


Netflix: Optimizing the user experience and investment decisions using Machine Learning

4 thoughts on “Knewton Personalizes Learning with the Power of AI

  1. Wow — fascinating article!

    Personalized learning is such a hotly debated topic — I’m curious to hear your thoughts on the logical conclusion of its place in the classroom. It’s hard to imagine school becoming a completely individualized experience, with students only learning from what is on the screen in front of them. It seems like there is value to group teaching and education (here at HBS with the case method we certainly think so!). So what do you think the role of personalized learning is actually going to be in the classroom?

    1. The answer to this is certainly up for debate. My sense is that adaptive and personalized learning become tools to supplement teachers and redesign what that role looks like, but not a complete substitute for human instruction. Essentially, picture 50 students working through their adaptive courseware individually, where the professor gets notified when someone is struggling and intervenes 1:1 in more of a tutor / coach role than in a lecturer role – a few schools are already doing this for some courses like math (GSU and ACC are the biggest I know of).

    2. @Daniel, I do not think that you can compare HBS to K-12 education. HBS selects for very similar specific profiles that have the same educational/ professional experience and motivation in a way and that understands what they are getting into vs. K-12 education is required from all students regardless of their needs and interests and capabilities. On the other hand, the value of group teaching and education can definitely be unlocked through different routes (group projects and activities, limited sessions where all the class is there, etc.).

  2. Great article! I find the idea of personalized learning such an obvious solution to the traditional educational system, that to be honest is not serving everyone the same way. For students that are less capable in certain subjects, it becomes quite hard to catch up. And the expectation from the educational system to be one size fits all should also be adjusted to reflect the new reality that not everyone is interested in the same things/ capable of doing the same things. I think one key challenge that personalized education presents is how to assess students’ capabilities in a standard way, particularly relevant for high school students applying to universities – I guess you can still maintain standard testings such as the SAT and potentially other specific standard tests to be completed before joining a specific program (e.g. Math and Physics prior to joining an Engineering program).

Leave a comment