“AI Native 100”: A Lean Startup Sample Earning $6 Million Annually Created by a Gen Z Dormitory Team for Consumer AI “Pixar”
“AI Native 100” is an AI Native Innovation Special Column launched by HuXiu Technology Group, and this is the 19th piece in the series.
Starting with four people, all Gen Z, from a university dormitory, they created a business with annual revenues of $6 million; attracting over 5 million users in just two years, with social media views surpassing 500 million; achieving sustainable profitability within their first year of entrepreneurship, with stable monthly revenues exceeding $500,000 and profit margins maintained at 20%-30%—in the wave of AI entrepreneurship, the Oleve team has written a classic case of lean startup with a set of impressive data.
To define Oleve, it is not a single AI product but rather an “AI factory”: with rapid iteration to generate AI products at its core, relying on viral marketing to efficiently accumulate users; the founders have accurately hit the growth code of the AI era, forging a Eruption path for a small team.
Departing from the dormitory: Two bestsellers, two traffic miracles
The story begins in early 2023 at the University of Rochester dormitory in the United States. At that time, student Sid Bendre and three roommates jointly launched their first product — the math homework assistance app Quizard AI, with a very direct core function: users scan the problems, and they can get real-time AI tutoring.
To cold-start, they posted a short video on TikTok with a cleverly worded title: “ChatGPT and Photomath had a child.” Unexpectedly, the video exploded overnight, with views soaring to 2.3 million; within less than 30 hours, the app attracted 10,000 registered users. After user verification, the team quickly launched a paid model: a weekly subscription fee of $9.99, rapidly achieving a revenue closed loop.
After the initial victory, the team launched a second learning-related product, Unstuck AI, in August of the same year. This learning assistant has more comprehensive functions: it supports users to upload unstructured learning materials such as PDFs, PPTs, classroom recordings, and YouTube videos, automatically converting them into an interactive knowledge base; more importantly, every response given by the AI is labeled with its source, which not only solves the AI “hallucination” problem but also enhances the credibility of the answers.
This time, their growth strategy is more aggressive: using handwritten sticky notes as visual materials, paired with TikTok viral audio, and even all videos follow the background style of platform top influencer MrBeast; the team also went to various schools for street interviews, edited the content, and then published it.
The aggressive strategy brought unexpected returns: Unstuck AI launched 30 days ago, and its related TikTok video views exceeded 250 million; within 9 weeks, it added 8 million new users, once ranking fourth on the App Store’s education category list.
Growth secret: “Lean Game Manual” and AI automation
Oleve’s Eruption is not accidental, CTO Sidhant once dissected the team’s “Lean Game Manual” in a public video, with the core being four principles, which can be summarized after implementation as:
Attract the right people, not just more people: Reject “just filling numbers,” prioritize hiring cross-disciplinary Compound talent, such as marketing professionals who understand programming, engineers with a product mindset, ensuring everyone can handle multi-functional roles;
Profit-first mindset at the grassroots: Each team member is responsible for one key business metric, review and advance weekly, eliminating the “burn money for scale” model from the early stages of entrepreneurship;
Integrated workflow: Consolidate product development, marketing, user operations, and other processes onto a single platform to reduce efficiency losses from cross-tool collaboration;
Reject relearning: Once a strategy is proven viable, standardize and reuse it immediately. For example, the TikTok marketing logic for Quizard AI was directly applied to Unstuck AI, and further automation was implemented afterward.
The core implementation tool of this manual is the team’s in-depth application of AI Agents. They developed an “Internal Agent” that can capture trending TikTok audio 24/7, automatically generate multiple versions of marketing scripts, and complete A/B testing; even the judgment of product iteration directions relies on AI tools for assistance — the team doesn’t need to repeatedly test and fail but instead lets AI filter out high-probability successful paths.
This “AI-driven automation” approach is also reflected in early technology selection. In early 2023, when OpenAI had not yet opened its general-purpose large model API, the Oleve team used the “multi-account cycling call” method, leveraging the Codex model originally designed for programming, to support natural dialogue through prompt engineering. This not only significantly reduced computing costs but also secured a time window. Later, they became the highest-volume user of the OpenAI Codex model, which also laid the foundation for their subsequent shift to the paid GPT-3.5 API and achieving profitability.
The essence of the model: Pixar in AI consumption, relying on “middle platform + matrix” for replication success
Although both breakout products focus on the education sector, Oleve never defines itself as an “AI education company” — they prefer to call themselves “Pixar in the field of AI consumption”: just as Pixar relies on industrialized processes to continuously produce animated films, Oleve aims to continuously incubate AI consumer products for different scenarios through standardized middle platforms.
The core of this model is “Operational Middle Platform + Product Matrix”:
Middle Platform Level: Build a unified technical framework (such as Agent systems, Prompt template libraries, testing pipelines) and marketing system (such as TikTok automated marketing tools, user feedback analysis models) to solve the “reinventing the wheel” problem;
Matrix Level: Based on middle platform capabilities, quickly develop products for different target audiences and scenarios. In addition to the education sector, the team is already exploring new products in non-education fields, one of which went from development to profitability in just three weeks.
This “incubator-style” model has also gained recognition from capital and the industry:
Jiahuo Capital CEO Yuan Ziheng believes that the core value of Oleve lies in the “reusability of the growth methodology” — its platform model can quickly respond to the needs of different C-end scenarios, which is the underlying logic for the business model’s success, similar to “building a solid skeleton first, then filling it with different scenarios’ flesh and blood”;
Silicon Valley venture capital firm Up Honest Capital points out that Oleve has validated the new logic of entrepreneurship in the AI era: AI development tools lower the barrier for applications, at this point, “deep understanding of user needs + unique GTM (go-to-market) techniques” equals “tools with traffic and revenue,” allowing small teams to break through.
Insights for small teams: 3 key elements of lean startup in the AI era
Oleve’s story also provides a reference path for small team entrepreneurship in the AI era:
1. Organizational Structure: Lean and Agile, Rejecting “Bloat”
From an initial team of 3 in a dormitory, to a later core group of 4, and only expanding to 6 by 2025, Oleve has consistently maintained a very small team size. Internal roles are clearly defined yet flexible: they set up two types of roles, the “Harvester” (Product Engineers focused on user experience and product expansion) and the “Farmer” (AI Software Engineers focused on the middle platform and infrastructure). This covers core needs while avoiding personnel redundancy.
The founder mentioned that the team drew inspiration from Palantir’s organizational experience, aiming to create a “consumer version of Palantir” – leveraging efficient collaboration and cross-functional capabilities to move a large market with a small team.
2. Growth Strategy: Automation + Viral, Reducing “Labor Costs”
Oleve’s marketing growth has almost achieved “half-automation”: from hot material scraping and script generation to A/B testing, all completed by AI Agents, with the team only needing to focus on adjusting strategic directions. This model not only reduces labor costs but also allows for quick adaptation to the flow of social media traffic, such as the fast-paced trends on TikTok, where manual real-time responses are difficult, while AI can achieve “all-weather trend chasing.”
At the same time, their content strategy accurately hits the “user resonance points”: for example, using colloquial expressions like “ChatGPT and Photomath had a child,” to lower the barrier of user understanding; conducting street interviews on campus to get closer to the target audience (students), allowing the content to naturally form Virality.
3. Business Logic: First Disperse Trial and Error, Then Focus on Deep Cultivation
Oleve’s “matrix approach” is essentially an “efficient trial and error” strategy in the AI era: in the early stages, multiple products and scenarios are laid out to quickly validate user needs and identify the most promising directions; once a product model is proven, resources are concentrated to deepen its development, while successfully accumulated experience is replicated to other products.
This aligns with the “Designer Founder” concept proposed by YC (America’s largest incubator) — YC believes that excellent founders need to possess “user empathy, problem-solving skills, and high standards of quality,” and Oleve has achieved the leap from “dormitory entrepreneurship” to “annual revenue of 6 million US dollars” precisely by accurately grasping student needs, refining product details, and rapidly optimizing growth strategies.
As Yuan Ziheng said: “Looking across industries, P&G succeeded with its brand matrix, ByteDance succeeded with its app matrix, and in the current AI Agent field, the matrix approach is equally applicable — first cast a wide net to increase the probability of hitting the mark, then focus on core areas to build long-term advantages. This may be the shortcut for small teams to break through in the AI era.”
