ScrollEd Turns Textbooks Into a TikTok-Style Feed

ScrollEd, a Palo Alto startup, is competing in the Startup Battlefield 200 at TechCrunch Disrupt with an app that turns PDFs, textbooks and other text files into a TikTok-style feed. The company plans to launch its consumer business at Disrupt, scheduled for October 13-15 in downtown San Francisco. TechCrunch

The input is static, long-form text. The output is a scrollable sequence modeled on TikTok and Instagram Reels. As the user scrolls, the feed alternates between AI-generated video, audio, text cards and interactive quizzes.

Navigation follows short-video habits. Swiping up surfaces a new topic. Swiping sideways goes deeper into the current topic. A quiz closes the sequence.

The co-founders are Utsav Gupta and Rebecca Neff. They bootstrapped ScrollEd in 2026, meaning they built it without outside funding. Gupta studies AI and human purpose at Stanford, while Neff studies computer science at the University of Pennsylvania.

The plan covers three markets: schools, corporate training programs and individual learners. The model is freemium, with a free base product and paid upgrades. A free consumer feed sits below a paid ScrollEd Pro subscription, with annual institutional licenses as a third tier.

For institutions, the pitch is distribution plus telemetry, or usage tracking. Schools and employers pay to deliver material through ScrollEd. In return they receive reports on engagement, lesson progress and other learning data.

The 2026 Disrupt takes place at Moscone West. The Battlefield 200 cohort will compete for a $100,000 equity-free prize and the Disrupt Cup. Organizers framed this year's event around building an enduring company in the AI era, with more than 10,000 tech leaders expected. TechCrunch

The broader pattern here will be familiar to readers who follow media or business software. Every shift in interface brings startups that convert old content into the new container. Desktop publishing did it to print layouts. Mobile video did it to television. Generative multimodal models, AI systems that work with text, images, audio and video together, are now doing it to documents.

From a technical standpoint, the test for study material is stricter than it looks. Summarization, chunking text into short pieces, narration and quiz generation are tasks current AI handles well. Fidelity, keeping exact meaning, is harder. Textbooks depend on precise definitions, derivations, diagrams and cause-and-effect chains where a small omission changes the lesson. A feed built for fast swiping still has to show it can preserve that structure rather than flatten it into trivia.

In my view, the more interesting bet is not video generation. It is interaction state, how the app remembers where you are. A textbook has chapters and an index. A social feed has recency and relevance ranking. ScrollEd proposes a third structure, where sideways swipes encode depth and vertical swipes encode breadth, with assessment inline. If that mapping holds for learners, long documents become navigable without a table of contents. If it does not, usage may skew toward review and cramming rather than first-time learning.

Schools and employers should note one concern worth raising, the data exchange at the center of the model. Engagement traces, completion signals and quiz responses are useful for instructors and training teams. They are also sensitive. Buyers will want to know retention windows, how data is stripped of personal identifiers, and where AI processing runs relative to where student data is stored. ScrollEd has described what reports customers receive, not yet the controls around them.

On distribution, past experience suggests product alone will decide only half the outcome. Consumer learning apps often struggle to keep users past the first weeks, and institutional sales cycles run long. Starting with a free feed to build usage, then converting heavy users to Pro while selling site licenses to institutions, is a sensible hedge. It also means running two businesses at once, with different acquisition costs, support loads and success metrics.

Looking further out, the hopeful case is clear and plausible over time. Students already spend hours in vertical feeds. I have watched my own kids learn new tools that way first and read the manual later. If even a fraction of that feed time shifts to course material without a loss in comprehension, total time spent learning rises. That is a net gain even if the feed never replaces the textbook.