ScrollEd turns textbooks into a TikTok-style scrollable feed

The pitch

ScrollEd, a Palo Alto-based startup presenting at the TechCrunch Disrupt Startup Battlefield 200 competition, takes a contrarian position in the debate over shrinking attention spans: rather than fighting short-form habits, it wants to give them something substantive to consume. According to TechCrunch, the company has built an app that converts any text file — from dry PDFs to full textbooks — into a vertical feed that feels much like TikTok or Instagram Reels.

The premise, as the founders frame it, is simple: if students won't open a textbook, the textbook should adapt to the format they already live in.

How the feed works

Content in ScrollEd does not arrive as static pages. TechCrunch reports that as a user scrolls, material is rendered as AI-generated video, audio, text, or interactive quizzes. A swipe up moves to a new topic, while swiping sideways digs deeper into the current one, with a quiz waiting at the end of each thread.

That sideways gesture is central to the design. Co-founder Utsav Gupta told TechCrunch that a short video can spark curiosity, but pursuing that curiosity normally means searching for another explanation or leaving the app. The aim, he said, was to make going deeper as frictionless as moving on to the next clip.

Gupta also acknowledged the tension in borrowing social media mechanics for education, telling TechCrunch the company wants to "meet them where they are" while being deliberate and choosing not to maximise engagement on the platform.

Bootstrapped by two students

ScrollEd was bootstrapped this year by Gupta and his co-founder and wife, Rebecca Neff. Per TechCrunch, both are students — Gupta studies AI and human purpose at Stanford, while Neff studies computer science at the University of Pennsylvania — and the idea grew out of their shared dissatisfaction with the feeling of late-night doomscrolling.

The company is targeting three groups: educational institutions, corporate training programmes, and individuals who, in Gupta's words, "just want something better to scroll on." The consumer business launches at Disrupt, an event Gupta joked he hoped would let ScrollEd disrupt social media and short form.

Business model and roadmap

ScrollEd runs on a freemium model, according to TechCrunch: a free consumer feed, a paid ScrollEd Pro subscription, and annual institutional licences. Institutions pay to deliver educational material through the platform and, in exchange, receive reports on engagement, lesson progress and other data.

Over the coming months the startup plans to concentrate on its consumer launch, expanding its lesson library, building a source-checking workflow, and developing pilots for institutions. It is also working on infrastructure intended to eventually let educators assess individual learners and adapt material to their needs. Gupta described the long-term ambition to TechCrunch as "a social network that's better for you."

Why it matters

ScrollEd is a concrete test of whether the mechanics that made short-form video dominant can be repurposed for deep learning rather than distraction. Education publishers and edtech firms have spent years struggling to hold attention against entertainment apps; folding the textbook into the feed inverts that fight entirely. The sideways-swipe depth model and end-of-topic quizzes are a serious attempt to translate "engagement" into comprehension rather than watch time.

The model also raises hard questions. AI-generated video and text are only as reliable as their sources, which is presumably why a source-checking workflow sits on the roadmap. Institutions may welcome the engagement analytics, but there is no settled evidence that scrolling beats structured reading for retention. ScrollEd's stated refusal to optimise for engagement is a constraint that the platforms it imitates do not face — and a claim it will have to prove commercially. If it works, the startup offers a template for reformatting dense material for universities and corporate training alike; if it does not, it becomes another data point that depth and vertical feeds do not mix.