YZi Labs Leads $11M Seed in VideoTutor, Its First AI Software Bet
YZi Labs has led an $11 million seed round in VideoTutor, an AI education platform founded by 20-year-old Kai Zhao. It's the firm's first investment in AI software. Co-investors include JinQiu Fund (affiliated with ByteDance), Baidu Ventures, Amino Capital, and BridgeOne Capital.
For educators, this signals a push to make high-quality tutoring more accessible and to cut content prep time. VideoTutor converts any question into a personalized, animated, voice-guided lesson within minutes.
Key points for schools and districts
- First AI software investment by YZi Labs: $11M into VideoTutor's "AI Education Agent."
- Turns questions into animated lessons using an LLM plus a Manim-based rendering pipeline.
- 20,000 users and 20,000 videos generated in 10 days; 1,000 API requests from education platforms.
- Focus areas: K-12, SAT/ACT prep, STEM, and language learning.
- Funding supports R&D, infrastructure scale-up, and partnerships across North America and Europe.
How VideoTutor works (and why it matters)
VideoTutor pairs a large language model with a Manim-powered animation system to visualize equations, diagrams, and scientific concepts with precision. The company built a proprietary Layout Manager and a fault-tolerant LLM loop to reduce errors and speed up generation.
According to Dr. Jing Xiong, AI Investment Director at YZi Labs and former Google Gemini engineer, this hybrid approach delivers strong semantic accuracy and clear visuals at lower cost than diffusion-based video tools. For busy teachers, that means faster lesson assets and fewer retakes.
If you teach math, physics, or test prep, this is worth a pilot. The promise is simple: ask a question, get a coherent, animated mini-lesson you can assign or embed.
Learn more about Manim, the open-source math animation engine referenced here: Manim Community.
Equity angle: tutoring access and cost
Over 85% of U.S. students can't afford traditional tutoring, which often runs $60-$90 per hour. VideoTutor's model pushes the cost of an explainer toward near-zero once the system is set up, making personalized help available at scale.
This doesn't replace great teaching. It reduces the time from "student question" to "helpful explanation," especially for schools with limited staffing for after-hours support.
Where educators can apply this now
- Flipped and blended learning: Generate targeted explainers for pre-class homework and station rotations.
- Formative checks: Create quick, step-by-step videos tied to exit tickets or weekly skill gaps.
- IEP and MTSS support: Offer multiple representations (visual + narration) for concept reinforcement.
- Multilingual access: Provide explanations in students' preferred languages with consistent visuals.
- SAT/ACT practice: Convert tricky items into 1-3 minute micro-lessons students can replay.
How to pilot VideoTutor this semester
- Pick one unit with clear misconceptions (e.g., linear functions, stoichiometry, grammar cases).
- Generate 5-10 short videos for common misunderstandings. Keep each under 3 minutes.
- Deliver through your LMS and collect quick feedback: "Was this clear?" "What's still confusing?"
- Compare outcomes to last term: quiz scores, completion rates, and student confidence.
- Iterate weekly: refine prompts, pacing, and visuals based on student response.
Questions to ask the vendor before deployment
- Accuracy and review: What QA exists for math steps, definitions, and diagrams?
- Curriculum alignment: Can outputs map to your state standards or AP/IB frameworks?
- Data privacy: What student data is stored, for how long, and how is it protected?
- Accessibility: Are captions, transcripts, and screen-reader friendly versions available?
- Integrations: Does it support your LMS, SSO, and rostering (Clever/ClassLink)?
- Content governance: Can admins lock sources, templates, and grade-level constraints?
- Cost model: Per-seat, per-video, or site license-and what's the expected monthly usage?
What the funding means
With new capital, the team plans to enhance the animation engine, scale infrastructure, and partner with schools and universities across North America and Europe. Zhao noted they picked YZi Labs as lead due to a shared belief that education should be personalized, affordable, and accessible.
For districts, expect more reliable outputs, faster rendering, and enterprise features as the product matures.
Related: YZi Labs Leads $50M Seed for Better Payment Network (BPN)
YZi Labs also led a $50 million seed round for Better Payment Network, which is building stablecoin-based rails on BNB Chain to speed up cross-border settlements. The goal is to cut transfer times from days to hours and reduce fees from roughly 2% to 0.3% by removing pre-funded liquidity requirements.
BPN already supports corridors such as Brazil, Nigeria, Mexico, and the Eurozone, with plans for a DeFi FX market covering up to 20 regional stablecoins. YZi Labs said BPN's onchain liquidity network could outperform both legacy fintechs and fragmented Web3 payment systems, if adoption continues.
For education, the near-term impact is limited. Longer term, cheaper and faster cross-border payments could help international programs, vendor payments, and global EdTech billing.
Learn more about BNB Chain here: BNB Chain.
If you're upskilling in AI for the classroom
- Browse practical AI courses by role: Complete AI Training - Courses by Job
- See tools and courses for generative video workflows: Generative Video Tools
Your membership also unlocks:
