Dropped Out of Columbia to Build the First General-Purpose Learning Agent
March 2, 2026
At Columbia, Zhao Yilin and Jiang Ruohan, both newly enrolled, attended a quantum physics hackathon. The lectures spanned from logic gates to code implementation, and the two beginners struggled to keep up. On one side were impenetrable textbooks; on the other, ChatGPT's rapid yet shallow answers. But when they themselves broke down the core concepts layer by layer, they discovered it wasn't as complicated as they'd imagined. In that moment, they realized: knowledge isn't necessarily hard — what's hard is finding someone who can actually explain it clearly.
They weren't satisfied with surface-level answers.
Zhao and Jiang possess what they call "growth outside the structure" — an agency akin to that of an Agent. Not pulled along by predetermined paths, not merely completing assigned tasks, but moving forward with their own judgment. Initially, they studied at Columbia by day and built their company by night. Their meetings had almost no small talk; often the first words were simply: "What problem are you running into?"
In November 2025, Hyperknow, a general-purpose learning agent, was officially launched — built around real learning scenarios with personalized dedicated models, every answer traceable to its source. Three months later, Hyperknow 3.0 went live. It began to think and plan proactively, anticipating your next step.
From last September to now, Hyperknow has completed five major version iterations, with hundreds of minor updates.
Back in high school, Zhao Yilin often wondered: what would it be like to have someone beside him who truly understood his thinking, grasped his progress, and knew his complete background? Later, he gradually became that person himself — someone who could make hard problems simple. Eventually, he realized that what everyone was truly missing was a Hyperknow: a companion that understands your learning context and responds to every question.
From ChatGPT to True Understanding
01:08 Two Gen Z Columbia founders, dropping out to start up, launch globally after 5 product iterations 02:24 Proactive general-purpose learning agent Hyperknow 3.0 helps students make knowledge simpler 03:15 The biggest pain point in learning: no good channel that's both deep and accessible 05:41 "General-purpose" has two layers: universal use cases (problem-solving, reading papers, generating flashcards) + diverse model capabilities
Differentiation from ChatGPT
07:59 ChatGPT has Deep Research too — why pay for Hyperknow? 08:15 Personalization exclusive to learning scenarios 09:46 Hyperknow actively asks questions, clarifies needs, judges what you need most right now 11:14 Agent thinking searches 50 webpages, traces every paragraph 13:02 ChatGPT compresses reading; Hyperknow truly processes 1000-page documents
Xiaohongshu Cold Start and User Co-Creation
13:48 Most people want answers, not process; Hyperknow aims to provide explanations that go beyond the surface 16:07 With Hyperknow, knowledge that once required 16 years to learn could be absorbed in 3-5 years 17:58 First Xiaohongshu post went viral, first user group established, daily bug-fixing with users 20:39 The story of an 80-year-old user
Pricing and Team
22:42 Priced below ChatGPT, fully free tier available, affordable for students 23:33 The original intention: helping people learn better — not just business, but something good for humanity 36:44 The entrepreneurial catalyst: a quantum physics hackathon 37:30 Students learn autonomously in personalized ways; schools return to teaching people and values 45:33 Team members mostly converted from users — tried the product first, then decided to join 46:40 Agency, subjective initiative, growth outside the structure 47:24 Full product rebuild possible in one week; breakthrough progress measured in half-week units
AI-Era Education Transformation
54:41 What would be different if Hyperknow existed in high school? 56:39 Literature reading traces every sentence, every fact to its source — down to the specific page 57:36 Higher learning quality, shorter time — freeing up time to do what you truly want 01:00:46 Don't Ask Wikipedia, Ask Hyperknow 3.0 01:02:28 We believe the best learning products should be made by students and learners
This episode's shownotes were generated with assistance from Moonshot AI
Built around real learning scenarios with personalized dedicated models, Hyperknow 3.0 introduces the "All-Round Learning Assistant." It can connect to learning management systems like Canvas, helping users build their own course knowledge bases. The agent proactively captures changes in course progress, judges your likely next actions, and prepares before you even ask.
Hyperknow-Generated Cheatsheet
A cheatsheet is a single page of personal notes permitted in open-book exams. From the perspective of real students, Hyperknow divides the page into four columns, compressing margins and font size while preserving adjustable space. For STEM content, the system proactively extracts all core formulas, annotating the meaning of each symbol and variable.
Hyperknow-Generated Cellular Respiration Concept Explanation
Unlike general models like ChatGPT, which tend to compress source material, Hyperknow can process thousand-page documents and read sentence by sentence, ensuring every piece of information in generated explanations can be traced to its original source page.
Visit hyperknow.io to experience the latest product.
Want to share your thoughts? Hear the founders tell the real story behind it all? ZhenFund will host its first AMA in the podcast discussion group, inviting Hyperknow founders Zhao Yilin and Jiang Ruohan. Scan the QR code to add the ZhenFund assistant with the note "join group" — see you March 5 at 8:00 PM.
Executive Producers: Cindy, Menmen
Post-Production: Yanaego
This Is For Real is a general business podcast produced by ZhenFund, where the investment team shares the latest trends and industry insights with leaders across fields.
Founded in 2011, ZhenFund is one of China's earliest angel investment institutions. Since its inception, ZhenFund has actively sought out the most outstanding entrepreneurial teams and era-defining investment opportunities in artificial intelligence, chips and semiconductors, robotics and hardware, healthcare, enterprise services, new energy, cross-border expansion, consumer lifestyle, and beyond.
ZhenFund — your first stop in entrepreneurship!