Algorithms Over Content in Edtech
Oboe
The core bet here is that better diagnosis can beat bigger libraries. Riiid, QANDA, and similar startups win by getting a student to the next best question or explanation faster, not by owning thousands of full courses. That makes their product feel more like a live tutor that reacts to each mistake, and less like a giant shelf of lessons waiting to be searched.
-
Riiid built around adaptive test prep, then sold that intelligence beyond its own app. After raising $175 million from SoftBank Vision Fund 2, it pushed its AI and machine learning tools to education companies and schools, which fits a model where the algorithm itself is the product being licensed.
-
QANDA scaled on a narrow but high frequency workflow. A student points a phone camera at a math problem, OCR reads it, and the app returns a solution path. Reported scale of 90 million registered users shows how a single strong use case can create mass adoption without a broad content catalog.
-
Newer entrants like Alice.Tech follow the same pattern with generative AI. Students upload notes or slides, then the product turns that material into flashcards, quizzes, and study plans. Incumbents like Khan Academy and Duolingo are moving the other direction, layering AI tutors and faster content creation onto already massive libraries.
This part of edtech is heading toward a split between algorithm companies and distribution companies. Specialists will keep building sharper tutoring loops for exams, math, and study planning, while large platforms use their scale to bundle similar personalization into much broader products. The winners will be the ones that turn student interaction data into visibly better learning outcomes every week.