When to open:When you want to start a business / side hustle / evaluate others' projects / join a new company, and want to use a comprehensive framework instead of a single author's perspective. This wiki extracts consensus, explicit disagreements, and exclusive dimensions of each school from the 5 schools of methodology (yitang Truman's Five-Step Method / Wang Huiwen Tsinghua Product Course / Wang Kai Accounting School / Lao Huang Independent Products / Shensi Circle VC Perspective).
One-sentence core:There is no "right" entrepreneurship methodology, only a methodology that "matches your identity". Students / big-tech departing entrepreneurs / product managers / engineers / independent developers each take different paths.The biggest anti-pattern is using methodology as a BP tool — it should be used as an "assumption checklist".
0. Entrepreneurship Views of the 5 Schools
| School | Source | Core Question | One-Sentence Stance |
|---|---|---|---|
| Yitang Truman School / Key Assumption Scientific School | Yitang Opportunity Pre-Judgment Methodology / Yitang / Complete Five-Step Method System (Key Assumptions + Business Milestones + 5 Advancements of the Five-Step Method + Implementation) / Yitang / Advanced Business Pre-Judgment (Industry Canvas + Concentration + Cases) | "How to use a set of scientific methods to raise the success rate from 10% to 76%?" | "Use assumptions to drive business; the Five-Step Method is the container for key assumptions" |
| Wang Huiwen Tsinghua Product Course School / Meituan School | Yitang / Complete 66-Page Notes of Wang Huiwen's Tsinghua Product Course(Compiled by Patrick / 2020) | "Why is failure 30 times more common than success?" | "Scale effects are the gravity of the business world / True core competitiveness = spotting opportunities + continuous learning / Luck accounts for 90% of success" |
| Wang Kai Accounting School / Small and Micro Teams in the AI Era | Wang Kai's AI-Era Entrepreneurship Methodology and Business Judgment Framework | "How to do business to earn cash flow, instead of fundraising?" | "The leftover poison of the internet is that fundraising = success; we want to do real business / run multiple projects in parallel / do the math before taking action" |
| Lao Huang Independent Product School / Non-Technical Global Merchant | newtype · Vibe Coding Independent Product Practice / newtype Mental Method · Super-Individual Methodology | "How to build an independent product starting with 1 person + no technical background + small money?" | "Unit Economics + turn technical problems into business problems + use strengths as the skeleton" |
| Shensi Circle VC School / a16z Sequoia Perspective | Shensi Circle Author Complete Works Portal | "Is this AI company's valuation reasonable?" | "Look at how a16z / Sequoia / YC invest + whether the founder is reliable + company case library" |
Key Observation:The disagreements among the 5 schools are not a debate over right or wrong, but a debate over positioning:
- yitang teaches MBA students + entrepreneurs (scientific method system)
- Wang Huiwen teaches Tsinghua students (classic product methodology + Meituan DNA)
- Wang Kai writes for independent developers / small and micro AI teams (cash flow survival)
- Lao Huang writes for non-technical individuals (from 0 to independent product)
- Shensi Circle writes for PMs / investors who follow the AI industry (VC perspective)
1. Cross-Author Consensus (6 Items)
1.1 "Failures far outnumber successes", not an exception but the default
| Author | Data/Statement |
|---|---|
| Wang Huiwen | "Failure is 30 times more common than success" / listed 5 failure-success pairs (Nokia / iPhone — MPMan / iPod — Kik / WeChat — Smart ABC / Sogou — IE / Chrome) |
| Wang Kai | "Only one successful listed company comes out of thousands of projects; out of a base of 50,000 people, fewer than 100 can pocket a million" / "Projects will most likely die, 90% don't make much money" |
| yitang Truman | "Innovative pursuits of top positions < 3%, even 1‰" / business type success rate grading (profitable side hustle < 50% → small restaurant < 30% → independent scaling < 15% → innovative pursuit of top position < 3%) |
| Lao Huang | "Projects will most likely fail / don't give independent development 1 year of patience" |
| Wang Xing (recounted by Wang Huiwen) | "Win first, then seek battle(The Art of War) / don't find a beachhead you can't win" |
Consensus Conclusion:The value of pre-judgment and preparation lies in turning "low-probability events" into "high-probability viability".
1.2 Do the math before taking action; this is the entry ticket to scientific entrepreneurship
| School | Accounting Focus | Key Numbers/Tools |
|---|---|---|
| Wang Kai(Three Things) | Traffic acquisition cost / monetization efficiency / long-term operation and maintenance cost | "Single technology cost calculation:Full-stack 30,000 monthly salary → requires 13,334 monthly paying users / must do slightly larger projects" |
| yitang Truman(Four-Character Formula: Chai-Tui-Ping-Suan [Break Down-Deduce-Evaluate-Calculate]) | Demand frequency + scale + urgency → unit economics → ceiling | "Without quantification, there is no real demand assessment" |
| Lao Huang(3 Unit Economics Mental Methods) | Weigh your capabilities and resources / turn technical problems into business problems / start from your strengths | "Don't touch large/complex projects / customers don't pay for pretty code" |
| Wang Huiwen(Meituan DNA) | After sales/user volume is large enough, (compared with small competitors) it generates user experience advantages + cost advantages | "Scale effects are the gravity of the business world" |
| Shensi Circle | a16z / Sequoia's "25× revenue multiple + 5× burn multiple" + Sequoia Agent Economy three stages | VC perspective |
Consensus Conclusion:Incorrect accounting is the number one cause of death for AI products(Wang Kai personally experienced 6 projects, 5 of which died due to accounting problems).
1.3 Pre-judgment > Execution (Option Pool / Endgame / Key Assumptions)
| School | Statement |
|---|---|
| yitang Truman(Yitang Opportunity Pre-Judgment Methodology) | "Pre-judgment is often several times more important than Yitang's Five-Step Method / P-type vs L-type success rate differs by 100 times" |
| Wang Huiwen | "The internet is not involution; it goes through 3 stages: market-driven → leadership-driven → innovation-driven, and now it is the end of the market-driven stage so everyone feels it's hard" / "True core competitiveness = the ability to spot opportunities + the ability to keep learning" |
| Wang Kai | "For new products, spend 90% of energy on referral traffic; SEO is a lagging indicator" / one of 6 mental methods: "Don't challenge giants on their main battlefield" |
| Lao Huang | "Enter from a business angle, can you turn technical problems into business problems? Wholesale APIs then retail them" |
Consensus Conclusion:Pre-judgment in the 4 weeks to 4 months before launching is far more important than execution after launching.
1.4 Five-Step Method / Five-Dimension Framework (almost all schools use a similar structure)
Although named differently, the 5 schools are highly consistent on "what to break down for from 0 to 1":
| Dimension | yitang Five-Step Method | Wang Huiwen Product Course | Wang Kai Accounting | Lao Huang |
|---|---|---|---|---|
| Demand | ① Demand (Four-Character Formula: Chai-Tui-Ping-Suan [Break Down-Deduce-Evaluate-Calculate]) | 4 major traits of demand + supply-demand relationship | Find people who need to pay | Customers buy "accurately hit demand points" |
| Solution | ② Solution (Product Core + Unit Model) | Failure vs Success 30 times | Don't challenge giants on their main battlefield | Turn technical problems into business ones |
| Business Model | ③ Business Model (Unit Economics) | Scale effect ABC categories + network effect | Three accounting things + ARPU vs instant payment | Unit Economics |
| Growth | ④ Growth (LTV / Customer Acquisition Cost) | Diffusion of innovations + 4P + STP + high-frequency beats low-frequency | Traffic dividend evolution + SEO/SEM offense and defense | Use strengths as the skeleton (marketing/operations/product) |
| Moat | ⑤ Moat (Project Moat) | Porter's 5 Forces + 3 competitive strategies + Meituan platform strategy | Three steps for AI application layer (enter with AI → accumulate scale → extend non-AI moats) | Strengths + cross-platform configuration + business decisions |
Consensus Conclusion:These 5 dimensions are all indispensable; any missing one is a "fake move".
1.5 Replace "I think" with "assumptions" (Truman's soul-searching questions are inherited by multiple schools)
Yitang / Complete Five-Step Method System (Key Assumptions + Business Milestones + 5 Advancements of the Five-Step Method + Implementation) §III:Ω Model (Propose Assumptions → Validate Assumptions → Iterate/Eliminate).
5 soul-searching questions (any entrepreneurship discussion should go through these):
- Behind your "I think",what are the actual objective assumptions??
- How do you plan to validate this assumption??
- If the assumption does not hold, what will happen to your business?
- How much cost is needed to validate this assumption??
- Is this assumption a 60% confidence level or 80%?
Communication Upgrade:"我认为..." → "我的假设是 X / 验证方式 Y / 现状 Z".
Wang Kai also agrees: "The number one cause of death for AI products is ostrich mentality — burying your head in building without thinking about traffic, payment, and expansion"; Lao Huang also agrees: "Think clearly about business first, Cursor will help you write it."
1.6 You will definitely die, so you need to "run multiple projects in parallel / run multiple assumptions in parallel"
| School | Statement |
|---|---|
| yitang(76% Probability Math) | "Test 4 times a year → total success rate P = 1 - 0.9⁴ ≈ 34%" |
| Wang Kai(Parallel Multi-Project Operation) | "The probability of succeeding at 5 small projects in a row is lower than succeeding at 1 slightly larger project" / "After finishing a project that makes money and has low maintenance cost, do the next one; don't grind on low-ROI opportunities" |
| Wang Huiwen | "At first we only wanted to build a $1 billion market cap, 60-person company, we didn't expect it to grow that big later" |
| Lao Huang | newtype.pro + Prompt House + LearnFromYouTube + Super Analyst + newtype Profile + CLI + OS,6+ products in parallel |
Consensus Conclusion:Betting everything on one project = betting on a low-probability event; running multiple projects in parallel = pushing low probability to high probability.
2. Explicit Disagreements (4 Items — Do Not Smooth Over)
2.1 Disagreement 1: To fundraise or not?
| Perspective | Stance | Key Arguments |
|---|---|---|
| Wang Kai(Anti-Fundraising) | "Doing business is more important than fundraising" | "In the internet era, very few founding teams actually made money; on the contrary, business owners with annual profits of several million can still travel and play Texas Hold'em" / "Successful fundraising only gets you to the card table; whether you have a bad hand hasn't been verified yet" / "Several VC friends said the first choice for AI entrepreneurship is Beijing, because all talent is in Beijing → I rebut: the default premise is big business" |
| yitang Truman(Look at the Endgame) | "Secure A, strive for B" | 7-layer spectrum map / choose an interval not a point / Endgame 1-2 is side hustle (pocket money) → Endgame 7 is IPO (billion-level+), fundraising is only needed for Endgame 4+ |
| Wang Huiwen(Meituan No.2) | Fundraising is a means not an end | "At first we only wanted to build a $1 billion market cap, 60-person company, we didn't expect it to grow that big later" / Luck accounts for 90% of success |
| Lao Huang(Independent Product) | "Wholesale APIs then retail / 1 person + start with small money" | Prompt House Mac App $9.98 / collect money directly via Stripe / no fundraising |
| Shensi Circle(VC Perspective) | Focus on the rationality of fundraising itself | a16z $46B sovereign fund / "King-making strategy doubles valuation in 27-60 days" / Sequoia Agent Economy three stages / 25× revenue multiple |
Real Disagreement:This is not a philosophical debate, it's a positioning debate:
- Innovative type + want to build a big company (L4+) + have a team → look at yitang 7-layer spectrum + Shensi Circle a16z perspective
- Small and micro teams / AI applications → Wang Kai anti-fundraising route
- 1 person + independent product → Lao Huang route
- Focus on AI industry landscape → Shensi Circle VC perspective
2.2 Disagreement 2: Scientific School (Assumption-Driven) vs Experiential School (Intuition/Courage-Driven)
| Perspective | Stance | Statement |
|---|---|---|
| yitang Truman(Scientific School Extreme) | "Use assumptions to drive business / the Five-Step Method without key assumptions is just an empty shell" | Ω Model / 5 soul-searching questions / replace "I think" with "assumptions" |
| Wang Huiwen / Wang Xing(Experiential School) | "Courage(Wang Xing answering China Mobile: what is core competitiveness)" + "Luck accounts for 90% of success" | "Xiaonei went against the mainstream to do acquaintance social networking, relying on the observation that 'users visit once every half month' + intuitive judgment" |
| Wang Kai(Practical School) | "Do business accounting + run multiple projects in parallel + take advantage of dividend periods" | 6 mental methods are practice-oriented; assumption thinking is implicit but not emphasized |
| Lao Huang("Just Do It") | "JUST DO IT, don't wait until you finish learning to start" | "Whether you are making content or building products, you absolutely should not wait until you have learned everything to start. Everything gets solved as you move forward." |
| Xiaohongshu vibecoding | "Aesthetics first, the model does not matter" | Intuition-driven + replication + fine-tuning |
Genuine disagreement:This is the fundamental split in entrepreneurial philosophy:
- First-time founder / high risk aversion / insufficient capital → Truman scientific school (systematically reduce risk)
- Highly experienced / already has directional sense / strong intuition → Wang Huiwen experience school
- Rapid iteration period of the AI tool stack → Lao Huang Just Do It (execution cost is extremely low, the cost of assumptions rises relatively)
Key insight (Truman's own response):P-type vs L-type differ by 100x;But the L-type does not eliminate the passion of the P-type; instead, it adds research and subtraction to the P-type.
2.3 Disagreement 3: How far should you take the "addition" of the option pool?
| Perspective | Goal of addition |
|---|---|
| yitang Truman(Yitang opportunity pre-judgment methodology §2) | "A general business needs at least 10 options / training-state MBA camp targets 100 options" |
| Wang Kai(Firsthand product list) | 11 firsthand projects + lessons / does not emphasize "addition" but instead emphasizes "rapid stop-loss" |
| Wang Huiwen(Scale effects) | Does not give a direct number / emphasizes "finding a temporary blue ocean in a huge market" |
| Lao Huang(Independent product timeline) | 6+ products in parallel(newtype.pro / Prompt House / LearnFromYouTube / Super Analyst / Profile / CLI / OS) / each iteration builds the foundation for the next |
Genuine disagreement:The goals of addition are different:
- Truman:Single business direction Do addition (find the best entry point)
- Wang Kai:Multiple businesses in parallel Do addition (find the best money-making opportunity)
- Lao Huang:Add new features to every product through iteration(find the best product form)
2.4 Disagreement 4: Domestic vs overseas
| Perspective | Stance |
|---|---|
| Wang Kai | "The domestic market is hyper-competitive; overseas is a high ground that must be entered" —— domestic competition gets fiercer by the day / after PMF overseas, subscription rate is 3-5x / single-user ARPU is 2-3x higher |
| yitang Truman | Domestic-focused(MBA students + private board members mostly use domestic cases / milk tea shop in Qingdao / Yitang's own business) |
| Wang Huiwen | Domestic-focused(Xiaonei / Meituan) / but noted "Uber lost to China/Russia/Indonesia; these places have large population bases + high education levels, and fierce competition instead improves the international competitiveness of Chinese companies" |
| Lao Huang | Overseas independent products(North American users / Stripe / tax filing compliance) |
| Shensi Circle | Focuses on the US AI industry(a16z / Sequoia / OpenAI / Anthropic / Cursor) |
| Lao D | Domestic adaptation(DeepSeek + Douyin + Xiaohongshu) |
Genuine disagreement:This is a comprehensive judgment of geography + user group + product type:
- AI application layer + too competitive to survive → Wang Kai going-global route
- Traditional industries (F&B / education / manufacturing) → yitang / Wang Huiwen domestic route
- Solo non-technical independent product → Lao Huang North America route
- Focus on the US AI industry → Shensi Circle
3. Core of the yitang Truman school (the most complete methodology)
3.1 76% probability math (the most persuasive point)
拍脑袋选方向 → 一次成功率 10%
↓
调研后排 90% 不靠谱 → 一次成功率 30%
↓
一年测 4 次 → 总成功率 P = 1 - 0.7⁴ = 76%
Turn low-probability events into things that have a high probability of working out.
3.2 15-character core method + 9-character mindset
15-character core method:Proactively think about the endgame, fully do addition, professionally do subtraction.
9-character underlying mindset(corresponding to timing, location, and people):
| Dimension | Objective aspect | Mindset |
|---|---|---|
| Timing(industry opportunity) | Not up to you, e.g., industry cycles / capital fads | Don't force it(go with the trend) |
| Location(business model) | Objective laws, e.g., unit economics / ceiling | Don't be naive(rapid validation) |
| People(original intention/team/resources) | Controllable, e.g., team capability / cooperation | Don't be afraid(proactively change) |
3.3 7-layer endgame spectrum diagram
graph LR
L1["1 副业<br/>极低风险<br/>零花钱"] --> L2["2 个体<br/>低风险<br/>养活自己"]
L2 --> L3["3 单体<br/>中风险<br/>几十-几百万"]
L3 --> L4["4 小规模<br/>中高风险<br/>百万级"]
L4 --> L5["5 体系增长<br/>高风险<br/>千万级"]
L5 --> L6["6 细分头部<br/>极高风险<br/>亿级"]
L6 --> L7["7 做大上市<br/>极高风险<br/>十亿级+"]
classDef low fill:#e8f5e9,stroke:#43a047
classDef mid fill:#fffde7,stroke:#fbc02d
classDef high fill:#ffebee,stroke:#e53935
class L1,L2 low
class L3,L4 mid
class L5,L6,L7 high
Rule: choose an interval not a point — "secure A while striving for B" (e.g., secure 3 while striving for 5).
3.4 Five-step method 100-version iteration path
| Version | Theme | Focus |
|---|---|---|
| V1-V10 | Launch and exploration | Bold assumptions + rapid research and validation |
| V10-V30 | Core refinement | Find the precise product core + make the unit economics work |
| V30-V50 | Business model | Unit economics baseline + dynamic forecasting |
| V50-V100 | Growth + moat | Growth engine + long-term moat building |
See Yitang / Complete five-step method system (key assumptions + business milestones + five-step method 5 progressions + implementation) / Lean startup / MVP / key assumptions (cross-author thematic synthesis) / Opportunity pre-judgment / business pre-judgment / endgame perspective (cross-author thematic synthesis).
4. Core of Wang Huiwen's Tsinghua product class school (classic legacy)
4.1 5 must-read books (personally selected by Wang Huiwen)
| Book | Value |
|---|---|
| The Paypal Wars | The author actually worked there / Elon Musk is portrayed negatively in the book |
| The Tipping Point | Vivid examples, explains product features + why iterations are accepted by consumers |
| The Lean Startup | Make validation faster and costs lower / Shorter iteration cycles than others = huge competitive advantage |
| The Innovator's Dilemma | Companies disrupted in the hard drive industry It was not because of technology, but because of organizational decisions |
| The Philosophy of Retailing (Suzuki Toshifumi) | Can be read from beginner to advanced level / covers 7-Eleven's decision-making process |
Not recommended:“Nine Failures One Victory”(Meituan biography) + “Tencent Biography”(both are promotional works).
4.2 5 PM greats + Wang Xing quotes
| PM | Value |
|---|---|
| Neil McElroy(Procter & Gamble, economics major) | The first product manager in human history / later served as P&G president + US Secretary of Defense |
| Steve Jobs | PC + graphical interface + iPhone(At the cutting edge of innovation you must learn the most cutting-edge theories) |
| Zhang Xiaolong | Foxmail + QQ Mail + WeChat (domestic product great from an engineering background) |
| Pichai | Chrome (metallurgical engineering major / Promoted from the Chrome business line rather than search) |
| Mark Zuckerberg | Facebook(All built social networking sites, cognitive accumulation) |
Core: product management has nothing to do with major,it is related to the person's worldview/values/way of thinking.
Wang Xing answering China Mobile's question "What is your core competitiveness?":"Courage".
4.3 The real 2 core competencies
- The ability to discover opportunities —— when real opportunities come, it is easy to produce results
- The ability to keep learning and improving —— Meituan went from social to group buying to food delivery, building up capabilities bit by bit
4.4 Scale effects = gravity of the business world
"After sales/user volume becomes large enough (compared to smaller-scale competitors), it produces (i)user experience advantage and (ii)cost advantage, depending on the specific business model."
See Unit economics / economies of scale / business model (cross-author thematic synthesis) / Yitang / Complete 66-page notes on Wang Huiwen's Tsinghua product class / Yitang / Complete product core + unit economics system (steps 2+3 of the five-step method).
4.5 High-frequency beats low-frequency (core method for entering new tracks)
When to use: when doing platform strategy / cross-category expansion, first do high-frequency categories to accumulate user base, then use high-frequency to drive low-frequency.
Meituan's "group buying → food delivery → hotel & travel → mobility" is a typical path of high-frequency beating low-frequency.
5. Core of Wang Kai's number-crunching school (small teams in the AI era)
5.1 6 entrepreneurship maxims (memorize 6 sentences and it's enough)
| # | Maxim | One sentence |
|---|---|---|
| 1 | Making money is more important than raising funding | In the internet era, very few founding teams actually make money |
| 2 | Calculate three things clearly | Customer acquisition cost / monetization efficiency / long-term operation and maintenance cost |
| 3 | Do not challenge giants on their main battlefield | Large model upgrades will not eat up many entrepreneurial opportunities; only hardware + systems + ecosystems combined with AI will |
| 4 | Find people who are willing to pay | Don't look for users, look for people willing to pay more than the customer acquisition cost |
| 5 | Run multiple projects in parallel | Domestic competition is on low-ROI opportunities, and the AI dividend window is short |
| 6 | Going global requires localization | Launch is 10x slower but the model is much better after PMF |
5.2 Number-crunching models
ARPU model vs instant payment model:
| Model | Applicable scenario | Focus |
|---|---|---|
| ARPU(calculate total value over an extended period) | Long-retention products; even a 3-year total value of 500 yuan is worth doing | Retention / renewal |
| Instant payment(every user must pay back immediately) | High-intent demand (writing papers / AI PPT) | Traffic arbitrage / single-transaction payback |
Wang Kai's lesson of losing 200 USD with 0 users: thought all high-ARPU models were suitable for instant payment.
Single technical cost calculation: full-stack engineer monthly salary 30,000 yuan → must do slightly larger projects; the probability of succeeding at 5 small projects in a row is lower than succeeding at 1 slightly larger project.
5.3 The "four big mountains" facing AI entrepreneurs
graph TD
A[小微 AI 创业者] --> B[硬件层<br/>苹果 / OPPO / 特斯拉]
A --> C[系统层<br/>iOS / Android / Windows]
A --> D[平台/生态层<br/>Chrome / 微信 / Slack / Notion]
A --> E[大模型层<br/>OpenAI / Anthropic / Google]
A --> F[大公司应用层<br/>Salesforce / Adobe / 美图]
Practical advice: don't do short Siri-path products / don't do things that phone manufacturers will definitely include in their systems / find niches that "hardware + systems + platforms + big companies" are all unwilling to touch.
5.4 Long-term "three-step" framework for the AI application layer
graph LR
A[Step 1: 用 AI 红利切入<br/>低门槛 + 快验证] --> B[Step 2: 获取流量 / 规模优势<br/>趁红利期攒护城河]
B --> C[Step 3: 延伸业务壁垒<br/>让 AI 占比降到 30% 以下<br/>构建非 AI 护城河]
Counterexample: simply connecting to an API + finding a niche scenario to solve one specific problem = no moat, eaten by the next version.
5.5 Wang Kai firsthand product list (11 projects + lessons)
| Project | Time | Result | Key lesson |
|---|---|---|---|
| enjoy food e-commerce | Before 2018 | Company experience | Raised 100 million USD in funding but the founding team did not make money |
| AI e-commerce shopping guide site | 2023 | Shut down | Early token costs could not be balanced |
| FastFacts(SEC scraping) | 2023 | Shut down | Carrying fee ceiling of 300,000 yuan / quant team was 5 seconds faster than us |
| MJ for Feishu | 2023 | Open source | Good traffic but not profitable |
| AiPathly(AI job hunting) | 2024 | Shut down | Innovative product doing SEO mismatch |
| Quanta Quest(C-end wiki AI search) | 2024 | Shut down | Blocked by data permissions |
| Squirrel Quick Look | 2024 | Continuously operating | Domestic C-end scale is limited but retention is good |
| AI smoking cessation product | 2025-2026 | Continuously operating | "Treat AI as a coach" |
| RSI AI investment strategy | 2025-2026 | 2-4% monthly return | See Wang Kai RSI AI investment strategy and the "good asset good price good volatility" framework |
| KeepRule(Investment principles wiki) | 2026 | Demo launched | AI Coding self-running database |
6. Core of Lao Huang's independent product school (non-technical solo route)
newtype · Vibe Coding independent product practice §1 Unit Economics 3 maxims for non-technical independent development:
| Approach | Content |
|---|---|
| Weigh your capabilities and resources | One person + non-technical + spending a little money → do not touch large/complex projects |
| Start from a business angle | A project does not need a technical moat to succeed.Can you turn technical problems into business problems/arithmetic problems? Wholesale APIs then retail them |
| Start from your strengths | If marketing, operations, and product are your strengths,then use these as the skeleton to architect the entire project, with technology as execution |
Why this is a moat for non-technical independent developers:
- Programmers' strength of "solving problems with the most elegant code" is overrated in independent development scenarios
- Customers do not pay because code is beautiful,they pay because "the need is hit accurately + the price is reasonable + the experience is smooth"
- The limitations of non-technical people instead force the discipline of "first think clearly about the business, then use AI to write code"
Lao Huang 6 maxims(borrowed from newtype maxims · super-individual methodology):
| Maxim | Content |
|---|---|
| Unfair Advantage three capabilities | "What do I have that others cannot copy?" |
| Just Do It | Don't wait until you finish learning to start |
| Treasure hunting vs fault-finding | Two basic modes of looking at things |
| Go Fundamental | Return to the base layer (Keynote replaces v0.dev for design) |
| How vs What | Emphasize How not What |
| 5 core values | Independent thinking / long-termism / first principles / continuous learning / creating value |
7. Core of Shensi Circle VC school (659 articles from a16z / Sequoia perspective)
Entry:Complete entry to Shensi Circle authors —— the deepest perspective on "VC + company actions" among Chinese self-media
Topic wiki(added 2026-05-18):
- Shensi Circle / AI entrepreneurship 0-1 cases + growth playbooks —— Complete PMF + 0-1 growth + GTM path(102+ entrepreneurship / growth cases)
- Shensi Circle / AI primary market VC perspective (a16z + Sequoia + YC) —— VC perspectives from a16z + Sequoia + YC
- Shensi Circle / AI Coding company cases (Cursor/Lovable/Bolt/Devin/Replit) —— Cursor / Lovable / Readdy.ai entrepreneurship cases
- Shensi Circle / Chinese entrepreneurs + going global (74 articles) —— Chinese + going-global entrepreneurs (74 articles)
7.1 a16z three paths to PMF (Jason Rosenthal 2026-05)
"More money only drags out a bad ending longer. PMF = a group of real users continuously using the product."
| Path | Applicable | Key |
|---|---|---|
| 1. Build alongside top-tier customers | Enterprise-grade / high compliance | Find the most demanding customers / customer requirements = your spec |
| 2. Stand in front of the exponential curve | New infrastructure / new protocols | Bet on the next wave of technology dividends |
| 3. Be your own first customer | Engineered products | dogfood / Cursor / Anthropic internal use |
7.2 Sequoia US: 4 Terrifying Questions
| # | Question | Positive signal |
|---|---|---|
| 1 | What right does my company have to exist? | Founder-market fit / willing to commit for 10 years |
| 2 | Do people actually care enough? | Cold start > relationship customers / customers' eyes light up |
| 3 | Can my product truly change behavior? | "Lightbulb moment" / sticky feature points |
| 4 | (See Shensi Circle / AI Primary Market VC Perspective (a16z + Sequoia + YC) §2.1) | — |
7.3 Landmark Growth Cases (Shensi Circle Breakdown)
| Case | Data |
|---|---|
| Lovable | 50 million ARR in 6 months / 1 million users in 10 months / 6 unconventional growth tactics |
| Perplexity | 450 million ARR in three and a half years / 5 major growth engines (doused with cold water by early investors) |
| Readdy.ai(Chinese founder) | Nearly 5 million ARR in 4 months / Lanhu + MasterGo team |
| 9 months from 0 to $100,000 monthly revenue | Solo founder / AI video |
7.4 Founder Mode Upgrades the "Design Engine"
"Founder Mode is no longer enough — In the AI era, founders must upgrade from 'building the engine' to 'designing the engine'"(Former CEO of AppSumo)
8. How to Choose — Methodology Paths by Identity
| Who you are | Primary method | Starting point |
|---|---|---|
| Student / first-time founder / risk-averse | Yitang Truman scientific school | 15-character mental model + 76% probability math + 7-layer spectrum map → option pool of 100 |
| Left big tech / want to build a large company / have a team | Wang Huiwen Tsinghua Product Course school | 5 must-read books + 5 PM greats + economies of scale ABC + high-frequency beats low-frequency |
| Small micro AI team / surviving on cash flow | Wang Kai accounting school | 6 mental models + three accounting matters + four big mountains + multi-project parallel operation |
| 1 person + non-technical + independent product | Lao Huang route | 3 Unit Economics mental models + convert technical problems into business ones + use strengths as the framework |
| Follow US AI industry / VC perspective | Shensi Circle VC school | Direct grep ~/kb/sources/learning/深思圈/ |
| Ordinary people in China without scientific internet access + self-media monetization | Lao D 13 lessons (AI Self-Media and Content Engineering (Cross-Author Thematic Synthesis)) | Douyin / Xiaohongshu + DeepSeek + Openrouter + digital human in 30 languages |
Universal minimum steps(for any identity):
- Think about the endgame(Yitang 7-layer spectrum): choose an interval, not a point / secure A while striving for B
- Do addition(option pool): at least 10, 100 in training mode
- Professional subtraction: three hard flaw dimensions: business model / team / industry
- Do the math(Wang Kai three matters + Lao Huang Unit Economics): customer acquisition cost / monetization efficiency / long-term operation and maintenance cost
- Write a list of key assumptions(Truman Ω model): propose assumptions → validate → iterate
- Rapid validation(Wang Huiwen + Wang Kai + Lao Huang):A shorter iteration cycle than others = a huge competitive advantage
- Run multiple projects in parallel(Wang Kai): finish one, make it profitable and low-maintenance, then start the next
9. Citation List
Main original author wikis:
- Yitang Opportunity Prediction Methodology —— 15-character mental model / 76% math / 7-layer spectrum / 9-character mindset / P-type vs L-type / 4-stage deliverables
- Yitang / Complete Five-Step Method System (Key Assumptions + Business Milestones + 5 Advancements of the Five-Step Method + Implementation) —— Five-step method taken by 110,000 people / Key assumption Ω model / Version 1-100 iteration / L1-L5 ranks
- Yitang / Complete 66-page Notes on Wang Huiwen's Tsinghua Product Course —— 66 pages of Wang Huiwen's Tsinghua Product Course / 5 PM greats / 5 must-read books / economies of scale / Wang Xing quotes
- Yitang / Advanced Business Prediction (Industry Canvas + Concentration + Cases) —— Advanced business prediction / 5 dimensions of industry canvas / Zhou Zijing concentration CRn / 4 types of market evolution
- Wang Kai AI-Era Entrepreneurship Methodology and Business Judgment Framework —— Wang Kai 516 posts / 6 mental models / three accounting matters / four big mountains / list of 11 firsthand products
- Wang Kai AI Industry Judgment and Large Company Competitive Landscape Observations —— AI industry judgment / company competitive landscape / the three Scaling Law debates
- newtype · Vibe Coding Independent Product Practice —— Lao Huang independent products / 3 Unit Economics mental models / Web → Mac App / Go Fundamental
- newtype Mental Models · Super Individual Methodology —— 15 of Lao Huang's cognitions and mental models / Unfair Advantage / Just Do It / 5 values
- Shensi Circle Author Complete Works Portal —— 659 Shensi Circle articles / VC perspective / a16z / Sequoia / company case library portal
- dontbesilent AI Self-Media Course / 13-lesson Systematic Methodology —— Lao D's domestic 13 lessons / monetization for ordinary people
Related topic wikis (this series):
- Opportunity Prediction / Business Prediction / Endgame Perspective (Cross-Author Thematic Synthesis) —— In-depth expansion of opportunity prediction (Batch 7)
- Unit Economics / Economies of Scale / Business Model (Cross-Author Thematic Synthesis) —— Unit economics + economies of scale + business model (Batch 6)
- ROI / Decision Evaluation System (Cross-Author Thematic Synthesis) —— ROI decision system (Batch 6)
- Lean Startup / MVP / Key Assumptions (Cross-Author Thematic Synthesis) —— Lean startup / MVP / key assumptions (Batch 7)
- AI Investment and Valuation (Cross-Author Thematic Synthesis) —— AI investment valuation (Batch 4)
- AI Industry Judgment and Company Competition (Cross-Author Thematic Synthesis) —— AI industry judgment and company competition (Batch 2)
History
- 2026-05-18: First release of the Topic-dimension wiki. Synthesized and rewritten from 10 author wikis (Yitang Opportunity Prediction / Five-Step Method / Wang Huiwen Product Course / Wang Kai Entrepreneurship Methodology / Wang Kai Industry Judgment / Lao Huang Independent Products / Lao Huang Mental Methods / Shensi Circle Portal / Lao D Self-Media / Yitang Advanced Business Prediction). 4 disagreements are explicitly marked: attitude toward financing / scientific vs experiential / size of option pool / domestic vs overseas
Sources and related materials
- Yitang Opportunity Prediction Methodology
- Yitang / Complete Five-Step Method System (Key Assumptions + Business Milestones + 5 Advancements of the Five-Step Method + Implementation)
- Yitang / Complete 66-page Notes on Wang Huiwen's Tsinghua Product Course
- Yitang / Advanced Business Prediction (Industry Canvas + Concentration + Cases)
- Wang Kai AI-Era Entrepreneurship Methodology and Business Judgment Framework
- Wang Kai AI Industry Judgment and Large Company Competitive Landscape Observations
- newtype · Vibe Coding Independent Product Practice
- newtype Mental Methods · Super Individual Methodology
- Shensi Circle Author Complete Works Portal
- Shensi Circle / AI Startup 0-1 Cases + Growth Playbooks
- Shensi Circle / AI Primary Market VC Perspective (a16z + Sequoia + YC)
- dontbesilent AI Self-Media Course / 13-lesson Systematic Methodology
- Opportunity Prediction / Business Prediction / Endgame Perspective (Cross-Author Thematic Synthesis)
- Unit Economics / Economies of Scale / Business Model (Cross-Author Thematic Synthesis)
- ROI / Decision Evaluation System (Cross-Author Thematic Synthesis)
- Lean Startup / MVP / Key Assumptions (Cross-Author Thematic Synthesis)
- AI Investment and Valuation (Cross-Author Thematic Synthesis)
10. 42 Chapters Intertemporal Supplement: Let Direction, Organization, and Business Evidence Constrain Each Other
Intertemporally organized and supplemented on 2026-10-18. The following synthesizes interviews from 2023–2026; it does not backdate later understandings to earlier periods, nor equate financing, valuation, rankings, or number of prototypes with operational success. Self-reported figures from each period are unaudited; deductions and transferable methods are editorial syntheses, with specific sources linked to the original program texts.
10.1 Choose important problems, then inventory controllable conditions
In a 2024 PMF interview, Albert recommended seeking markets where user expectations are clear, current capabilities can deliver, multiple players exist, and latecomer paths can be studied. He said "do things without barriers," which does not mean never accumulating advantages, but rather not betting the first step of entrepreneurship on things that exist only in one's own imagination, with no evidence of demand or feasibility. External conditions include user jobs and substitutes, distribution, and payment habits; internal conditions include actual capabilities, cash, and tolerable trial-and-error. Being able to estimate the probability of success is different from being able to afford the loss; subjective confidence cannot be packaged as actuarial calculation.Original program text
By March 2026, Albert further separated "Why choose important problems" from "How to get things done well." He did not argue that one can only do familiar small things: one can attack new directions, but first find controllable variables and gradually build advantages, rather than relying on competitors happening to fail. Qu Kai places more emphasis on preemptively packaging technology and acquiring data and capital resources; Albert presses on what "winning" actually means—financing, valuation, making money, or users continuing to need the product. The disagreement was not resolved: preemption can improve conditions for the next step, but it cannot automatically prove that sustainable value has been created. For already profitable businesses, it may also be reasonable to stop chasing every technology hotspot; when old organizations cannot support new mechanisms, building a separate small closed loop is more feasible than forcefully overhauling the entire company.Original program text
In practice, project initiation can be written in three layers: first, what problem is worth solving, and what users will do if they do not use us; second, what we already control, and what we are still betting on; third, if the bet is wrong, when we will see it, and how much cost we can bear. Do not read "direction is more important than execution" as meaning no delivery is needed, nor read "rapid trial and error" as meaning problem judgment can be skipped. An experiment that finds attribution errors and corrects them also counts as valid positive feedback; morale does not need to be sustained by pretty results.
10.2 The same person's views change, and product stages also change
Albert's intertemporal changes are especially worth preserving. In 2024, he emphasized clear delivery, payment, and sustainable assets, and rejected one-off fad tools; in March 2026, he reflected on the appeal of 80-point AI entertainment content versus 90-point mature content, arguing that model novelty should not be a reason for users to pay an "AI tax." By May 2026, his team had built dozens of internal products in two months, yet admitted that revenue was zero at the time and they had not yet met external product standards; the direction shifted toward "resonance" among makers—being understood, collaborated with, and re-created by people who care.Original program text Original program text Original program text
This is not stitching all views into a single line of "always persist in PMF." The later episode places more weight on impact, experience, and connection, but he also distinguishes personal free creation from companies with payroll and shareholder responsibilities. Response is being seen; resonance is being understood by relevant people; reward is economic income; the three can reinforce each other but cannot replace one another. For creator platforms, the number of prototypes, likes, and revenue should be recorded separately; for companion products, the subjective experience of being understood must be cross-checked with repeat visits, long-term satisfaction, and safety feedback. Different product types can have different business validation chains, but a company must know which link in its chain is not yet closed.
Chen Yu's 2023 technology investment experience provides another kind of continuity: when Zilliz moved from GPU databases to vector databases, it did not throw away all old accumulations, but retained computing, database, and implementation capabilities, and confirmed the pivot using customers' new demand for image search; ease of prototyping does not mean ease of performance and global commercialization. The transferable method is to clearly list the capabilities retained during the pivot, the assumptions discarded, and the new customer evidence added, rather than restarting from scratch to chase new buzzwords. He also admitted that he did not predict the later impact of ChatGPT; professional understanding adds conditions for judgment, but does not grant the ability to foretell the future.Original program text
Mengqi’s 2026 reflection provides a more specific counterexample: because the general Agent space was crowded, she first chose the vertical track of “finding people,” then treated influencers, recruitment, experts, and customers as extensions of the same capability, only to eventually discover that shared retrieval technology does not equate to the same user task. Professional marketers want process control, while founders want direct results; when the target role is not selected, listening to every piece of feedback may drag the product toward mutually conflicting forms. When pivoting, record what facts disproved old assumptions and which segment of delivery the new expansion fills, rather than merely expanding the financing market narrative.Original program text
She also reflected that, due to self-identification and the financing story, she hired algorithm talent early and talked about reinforcement learning, later shifting energy to user intent, context switching, stability, and issue acceptance. The corresponding discipline is not “never hire algorithm engineers,” but that every technical investment must correspond to a currently measurable bottleneck. Anonymous peer accusations do not count as industry facts; that a Skill used internally at high density every day is more effective than the old product only demonstrates progress in an environment where data, permissions, and goals are aligned; selling it externally still requires addressing customer differences, authorization, and maintenance.
10.3 Timing is composed of multiple clocks, not determined by news heat
Zhou Jian observed in 2024 that models, budgets, client attention, computing power procurement, and application supply were out of sync. He believes opportunities arise from the intersection of these variables, and clearly distinguishes between clients asking you to provide free training and those who have already decided to trial. Clients buying GPUs first and then looking for applications may create a temporary window for computing power utilization management; he himself reminds that such opportunities may not last. Sunk expenditures cannot be directly regarded as sustained demand, nor can the 2024 prediction that “applications will explode” be treated as proof of subsequent actual growth.Original program text
Qu Kai emphasized the capital window in his 2024 Character.AI monologue: data improvement may not offset valuation multiple contraction, and financing progress and product progress are two different clocks. His deduction from transactions about model commoditization and that “the model war is nearly over” was a viewpoint at the time; a single transaction is insufficient to prove the industry endgame. His “fatalism” points out external constraints, but should not cause the team to abandon controllable actions. What is truly actionable is to set scenarios based on cash runway, procurement cycles, user maturity, and supplier dependence, rather than betting that the next round will definitely be financed at a higher valuation.Original program text
Jenny Xiao reminded in 2025 that a pretty high valuation also raises the revenue threshold required for the next round; when two or three customers contribute the majority of revenue, it sometimes resembles high-quality customized services and may not yet validate a scalable product. Only by separating market size, signed contracts, actual cash collection, standard module reuse, and renewals can one avoid covering five unvalidated problems with one good story.Original program text
10.4 Organizational methods must match the business, not copy the appearance of winners
Wei Xiaokang explains organizational differences using chain length, feedback cycles, and gross margins: high-margin, short-chain businesses are better able to find good talent first and then match them to positions; low-margin, long-chain businesses require more operational planning, resource budgeting, and collaboration precision. This coordinate system does not exhaust all organizations, but it helps judge what to learn. Even three people may have different understandings, so they should align on goals, responsibilities, and rolling plans; there is no need to first replicate the hierarchies, ranks, and lengthy performance materials of large companies.Original program text
Talent density should refer to people who match current tasks and can collaborate to solve difficult problems, not density of prestigious school graduates, high salaries, or a particular age group. For recruitment, first define the business problem and who is needed when, then build supply and outreach; beyond a single interview, one should also look at evidence of real work and permitted, reliable reference checks. Offers must clarify expectations on both sides and not replace fair compensation with grand meaning. Wei is relatively pessimistic about young founders and OPC, while Qu Kai points out that veterans may be constrained by rules and young people may open new paths; converting disagreements into verifiable resources, team composition, and task capabilities is more reliable than selecting people by birth year.
Zhou Jian’s experience growing organizations from ten to a thousand people supplements the “braking” after “takeoff”: early members will hit growth ceilings, and external managers may also not adapt to going from zero to one; old and new members, growth pace, cash, and responsibilities all need to be re-matched. Seniority does not equal loyalty, and old comrades are not naturally competent. The purpose of an organization is delivery, not batch elimination; growth in real tasks requires support and suitable battlefields, and should not become an excuse for “hire then filter out.”
Palona Ren Chuan’s 2025 practice divides AI teams by end merchant/consumer outcomes, with engineers directly contacting customers to reduce information loss across sales–PM–engineering. The approach for a team of about 20 people does not directly prove that product and governance roles are still unnecessary at 50 or 100 people; he says he does not know. Letting AI try first by default does not mean abolishing accountability. His experience with rapid review and rollback was challenged in Q&A by P0 issues, dirty data, and irreversible impacts, and he later acknowledged that Infra should be more conservative. Handoffs can be shortened, but production permissions, security reviews, and final acceptance should still be set independently according to risk.Original program text
Fan Haoyu provided another side of organization and product review in 2025: responsibility can be end-to-end, but “complete” cannot be understood as everyone being infinitely all-capable. Using the experience path from discovery, initial use, continuous exploration to upgrade and migration, he reminded teams not to treat improvement of a single link as complete delivery. User habits, existing content, and entitlements also form slow variables in software, and rapid revisions are not costless. Corresponding organizational design requires someone to be responsible for the entire user journey while retaining professional collaboration and risk gates.Original program text
He emphasized taste and overall feel; Qu Kai pressed that taste may be wrong or ahead of its time, and Fan also acknowledged that one must bear the results. This passage cannot be reduced to “data is useless,” nor can good data be used to eliminate all counterevidence from experience. A more practical combination is: write down what feels off as specific hypotheses, examine segments, long-term usage, failure cases, and actual verbatim quotes, then adjust implementation; “wanting both” means seeking designs that reduce conflict, not abolishing budgets and tradeoffs. Thus, an entrepreneur’s insistence on complete experience should coexist with judgments that can be overturned.
10.5 Growth is part of the product system, but sprints have boundaries
Joe of ACE Studio reviewed HF0 in 2025: teams that already had a paying foundation changed their way of working through 12 weeks of residency, a single goal, and weekly outcome feedback. The most dramatic doubling of daily revenue came from doubling prices, followed later by pricing models, KOL outreach SOPs, scaling winning video ads, model upgrades, and win-back of churned users. It was not belief that made cash appear out of thin air, nor can the final tenfold growth be attributed entirely to one trick. The account, attribution, conversion, and creative infrastructure in the first six weeks cannot be skipped.Original program text
What is reusable is putting functionality, pricing, outreach, and service into the same business loop and testing them one by one. A short-term revenue increase from a price hike still requires looking at subsequent refunds, renewals, and segment retention; self-reported ad ROI without complete gross margin and attribution frameworks cannot be directly used as a scaling parameter. A single sprint metric must be paired with quality and safety guardrails. After leaving the program, the team did not maintain extreme working hours; Joe instead switched to running R&D and growth sprints in separate cycles; limited high intensity does not equal a permanent work regime.
The proper place for belief here is to make the team willing to explore more candidate solutions and tolerate delayed feedback, rather than stopping the acknowledgment of bad news. Zhang Jinjian emphasized stable attention, difficult choices, and not turning setbacks into self-denial in a 2024 vitality interview; Qu Kai meanwhile argued that belief needs research support, and the two have a genuine difference on whether belief requires reasons. Expressions about frequency, energy, and physics are only analogies and cannot be treated as psychological causation or a science of success. For business decisions, affirmation of one’s own value can coexist with rejection of business assumptions: one need not deny oneself, but one must also allow the project to be falsified.Original program text
10.6 Resource-driven playbooks first ask about conditions; capital narratives and real-business results are verified separately
Resources are not original sin, nor are they a route every team can replicate. Zhou Jian’s delivery to central and state-owned enterprises relied on existing industry talent, commercial resources, and long sales chains; large-client computing power helped temper the product, but whether it can form high-margin standard modules still requires subsequent proof. The high-salary talent poaching discussed by Wei Xiaokang must have business returns and cash support; Albert criticized using ByteDance-style mass top talent and capital horse-racing as the default plan for ordinary teams, pointing out that good business and culture can also enable ordinary members to keep doing the right things. First clarify whether you are pursuing personal income, a sustainable small company, or venture-capital-style expansion, then choose your playbook.
For each project, retain at least four categories of evidence separately: what users get, what remains after the enterprise delivers, how money returns to the operating entity, and why capital is willing to set a price. Financing may improve supply and survival conditions but cannot replace renewals; an exit may bring returns to some shareholders but cannot prove the original product was profitable; a founder’s technical aspirations also cannot replace user tasks. The founder psychology and transaction motives in the Character.AI materials should especially retain the attribution of “Qu Kai’s inference” and cannot be packaged as inevitable destiny using hindsight results.
The minimum landing sheet for product teams is to revisit one page each month: the current most important user problem, three controllable conditions, three external assumptions, the next empirical test, cash and time caps, who is responsible for the outcome, and under what circumstances to stop. Team size and tool adoption are only means; if user value, business results, and personal goals conflict, write out the conflict and make tradeoffs rather than covering it over with a bigger story.
English version
10. A Synthesis Across 42 Zhangjing Episodes: Let Direction, Organization, and Business Evidence Constrain One Another
Added through a synthesis across episodes on 2026-10-18. The following draws on interviews from 2023–2026. It does not project later insights backward into earlier periods or equate funding, valuations, rankings, or prototype counts with business success. Self-reported figures from individual episodes are unaudited. Inferences and transferable methods are editorial synthesis; the source links point to the original episodes.
10.1 Choose an Important Problem, Then Assess the Conditions You Can Control
In his 2024 PMF interview, Albert recommended looking for markets where users have clear expectations, current capabilities can deliver, multiple players exist, and the paths taken by later entrants can be studied. His advice to “do things without barriers to entry” did not mean never building an advantage. It meant avoiding a first entrepreneurial step staked on something that exists only in one's imagination, without evidence of demand or feasibility. External conditions include user tasks and alternatives, distribution, and payment habits; internal conditions include actual capabilities, cash, and affordable experimentation. Estimating the probability of success and being able to absorb a loss are different matters. Subjective confidence must not be presented as actuarial precision. Original episode
By March 2026, Albert further separated “why: choose an important problem” from “how: do the work well.” He was not arguing that founders should pursue only small, familiar tasks. A team can enter a new direction, but should first identify controllable variables and gradually build an advantage, rather than rely on competitors happening to fail. Qu Kai placed more emphasis on being early to package technology and obtain data and capital resources. Albert asked what “winning” actually meant: funding, valuation, making money, or continuing user need. The disagreement remained unresolved: moving first can improve the conditions for the next step, but does not automatically demonstrate sustainable value. For an already profitable business, declining to chase every new technology trend may also be reasonable. When an existing organization cannot support a new mechanism, building a separate small, complete operating loop may be more feasible than forcing the entire company to change. Original episode
An actual project proposal can be written in three layers. First, what problem is worth solving, and what would users do without us? Second, what do we already control, and what are we still betting on? Third, if the bet is wrong, when will we know, and how much can we afford to lose? “Direction matters more than execution” does not excuse a failure to deliver; “experiment quickly” does not mean skipping judgment about the problem. An experiment that exposes and corrects faulty attribution also provides useful positive feedback. Morale need not depend on attractive results.
10.2 A Person's Views Can Change, and So Can a Product's Stage
Albert's changes across episodes are particularly worth preserving. In 2024, he emphasized clear delivery, payment, and sustainable assets, moving away from tools built around one-off waves of excitement. In March 2026, he questioned the appeal of AI entertainment content scoring 80 when mature content scored 90: model novelty should not make users pay an “AI tax.” By May 2026, his team had built dozens of internal products in two months, yet acknowledged that returns were zero at that point and that the products had not reached the standard for external release. The direction shifted toward “resonance” among makers: being understood by people who matter to them, collaborating, and creating further work from one another's creations. Original episode Original episode Original episode
These views should not be stitched together into “he consistently pursued PMF.” The later episode gave greater weight to impact, experience, and connection, while still distinguishing free personal creation from a company with salaries and obligations to shareholders. Response means being seen; resonance means being understood by relevant people; return means economic income. The three can reinforce one another, but cannot replace one another. A creator platform should record prototype counts, likes, and revenue separately. A companion product should cross-check the subjective experience of being understood against returning use, long-term satisfaction, and safety feedback. Different product types may have different chains of commercial validation, but a company must know which link remains incomplete.
Chen Yu's account of technology investment in 2023 offers a different form of continuity. As Zilliz moved from a GPU database toward a vector database, it did not discard all its earlier work. It retained capabilities in computation, databases, and implementation, and used new customer demand for image-to-image search to validate the pivot. An easy prototype did not imply easy performance engineering or global commercialization. The transferable method is to list the capabilities preserved in a pivot, the assumptions discarded, and the new customer evidence gained, rather than restarting from zero to chase a new term. He also acknowledged not predicting ChatGPT's subsequent impact. Domain expertise adds conditions for judgment; it does not grant foresight. Original episode
Mengqi's reflections in 2026 offer a more specific counterexample. Because general-purpose agents were crowded, the team first chose the vertical of “finding people,” then treated influencers, recruitment, experts, and customers as extensions of the same capability. It ultimately found that shared retrieval technology does not mean a shared user task. Professional marketers want control over the process; founders want the result delivered directly. Without choosing a target role, listening to every piece of feedback can pull a product into conflicting forms. A pivot should record which facts contradicted the old assumptions and which part of delivery the new expansion addresses, rather than merely enlarging the market narrative used for fundraising. Original episode
She also reflected on hiring algorithm specialists and discussing reinforcement learning prematurely because of self-image and the fundraising story, then later shifting effort to user intent, context switching, stability, and acceptance of completed work. The corresponding discipline is not “never hire algorithm specialists,” but to connect every technical investment to a currently measurable bottleneck. Anonymous accusations against peers are not treated as industry facts. That a Skill used intensively inside the company every day worked better than the old product only demonstrates progress in an environment with aligned data, permissions, and goals. External sales would still require addressing customer differences, authorization, and maintenance.
10.3 Timing Is Made Up of Several Clocks, Not News Attention
In 2024, Zhou Jian observed that models, budgets, customer attention, compute procurement, and application supply were out of sync. He saw opportunities at the intersection of those variables, while explicitly distinguishing a customer inviting you to give free training from one that had decided to run a trial. Customers buying GPUs before finding applications could create a temporary opening for compute-utilization management; he himself warned that such opportunities might not last. Sunk expenditure cannot simply be treated as continuing demand, nor can a 2024 prediction that “applications will take off” serve as evidence of growth that actually followed. Original episode
In his 2024 solo episode on Character.AI, Qu Kai emphasized the capital window: improvements in operating metrics might not offset shrinking valuation multiples. Fundraising progress and product progress run on different clocks. His transaction-based inferences about model commoditization and a “model war nearing its end” were views held at that time; a single transaction cannot establish an industry's final state. His “fatalism” highlights external constraints, but should not make teams abandon controllable actions. The actionable response is to build scenarios around cash runway, procurement cycles, user readiness, and supplier dependence, rather than betting that the next funding round will necessarily come at a higher valuation. Original episode
In 2025, Jenny Xiao warned that an attractive, high valuation also raises the revenue threshold needed for the next round. When two or three customers contribute most of the revenue, the business may resemble high-quality custom services rather than an already validated scalable product. Market size, signed contracts, cash actually collected, reuse of standardized modules, and renewals must be assessed separately to avoid covering five untested questions with one good story. Original episode
10.4 Match Organizational Methods to the Business, Rather Than Copying Winners' Outward Form
Wei Xiaokang explained organizational differences through value-chain length, feedback cycles, and gross margins. A high-margin business with a short chain is better positioned to hire strong people first and then match them to roles; a low-margin business with a long chain needs greater precision in operating plans, resource budgets, and coordination. These dimensions do not describe every organization, but can help determine what to learn from one. Even three people may understand a task differently; align goals, responsibilities, and rolling plans without first copying large-company hierarchies, job grades, and lengthy performance documents. Original episode
Talent density should mean people suited to current tasks who can solve difficult problems together, rather than a concentration of elite-school credentials, high salaries, or a particular age. Hiring starts by specifying the business problem and who is needed when, then building the candidate pool and outreach. Beyond one interview, examine evidence of real work and reliable background checks conducted with permission. An offer should clarify expectations on both sides; a grand sense of purpose does not replace fair compensation. Wei was relatively pessimistic about young founders and one-person companies (OPCs), while Qu Kai noted that veterans may be constrained by established rules and young people may find new paths. Translating this disagreement into testable resource, team-building, and task capabilities is more reliable than selecting people by birth year.
Zhou Jian's experience with organizations growing from ten people to a thousand adds the need for brakes after “takeoff.” Early employees may reach growth limits, while outside managers may struggle with zero-to-one work. New and existing members, growth pace, cash, and responsibilities all need to be matched again. Seniority does not equal loyalty, and longtime companions are not automatically qualified. An organization exists to deliver, not to eliminate people in batches. Growth through real work requires support and an appropriate setting; it should not become an excuse to “hire them and screen them out.”
In Palona's 2025 work, Ren Chuan organized AI teams around final outcomes for merchants and consumers. Engineers dealt directly with customers to reduce information loss across sales, PM, and engineering handoffs. What worked at approximately twenty people does not demonstrate that a team of fifty or a hundred still needs no product or governance roles; he said he did not know. Letting AI try first by default does not remove accountability. During the Q&A, his experience with rapid review and rollback was challenged over P0 incidents, dirty data, and irreversible effects. He later acknowledged a more conservative approach to infrastructure. Handoffs can be shortened, but production permissions, security review, and final acceptance should still be set independently according to risk. Original episode
In 2025, Fan Haoyu offered another side of organizational and product review. Responsibility can be end to end, but “complete” should not mean every individual is infinitely versatile. His experience path—from discovery and first use, through continued exploration, to migration across product generations—reminds teams not to mistake an improvement in one step for complete delivery. User habits, existing content, and entitlements also create slow-moving variables in software; rapid redesign is not cost-free. The corresponding organizational design needs someone accountable for the entire user journey while retaining specialist collaboration and risk gates. Original episode
He emphasized taste and the overall experience. Qu Kai asked whether taste might be wrong or ahead of its time, and Fan acknowledged the obligation to bear the consequences. This exchange cannot be reduced to “data is useless,” nor can good data erase all contrary evidence about experience. A more useful combination is to turn the feeling that something is wrong into a specific hypothesis, examine cohorts, long-term use, failure cases, and actual user quotations, then adjust the implementation. Wanting both sides of a tradeoff means seeking designs that reduce the conflict, not removing budgets and choices. A founder's commitment to the complete experience should therefore coexist with judgments that can be overturned.
10.5 Growth Is Part of the Product System, but Sprints Have Limits
In 2025, ACE Studio's Joe looked back on HF0: a team with an existing base of paying customers changed how it worked through a twelve-week residency, a single goal, and weekly outcome feedback. The most dramatic doubling of daily revenue came from doubling the price. Later work included pricing models, a standard operating procedure for KOL outreach, scaling winning video advertisements, model upgrades, and bringing back churned users. Belief did not conjure cash from nowhere, and the eventual tenfold growth cannot be attributed entirely to one move. The account, attribution, conversion, and creative infrastructure built during the first six weeks cannot be skipped. Original episode
The reusable approach is to place features, pricing, outreach, and service in one commercial loop and test them individually. A short-term revenue increase after raising prices still requires examining later refunds, renewals, and cohort retention. Self-reported advertising ROI without complete gross-margin and attribution definitions cannot be used directly as a scaling parameter. A single sprint metric needs quality and safety guardrails. After leaving the program, the team did not maintain extreme working hours. Joe instead alternated R&D and growth sprints by period; a limited spell of high intensity is not a permanent working system.
Belief has a reasonable role here: encouraging a team to explore more candidate approaches and tolerate delayed feedback, rather than stop acknowledging bad news. In his 2024 interview on vitality, Zhang Jinjian emphasized steady attention, difficult choices, and not turning setbacks into self-rejection. Qu Kai argued that belief needs research behind it; the two genuinely differed over whether belief needs reasons. References to frequency, energy, and physics are analogies, not psychological causation or a science of success. In business decisions, affirming one's own worth can coexist with rejecting a business hypothesis: a project must be allowed to be falsified without requiring a rejection of oneself. Original episode
10.6 Ask What a Resource-Intensive Strategy Requires, and Validate Capital Narratives Separately from Operating Results
Resources are neither inherently wrong nor a path every team can reproduce. Zhou Jian's delivery to central and other state-owned enterprises relied on existing industry talent, business resources, and long sales cycles. Large customers' compute resources helped refine the product, but whether that could produce high-margin standardized modules still needed later evidence. Wei Xiaokang's discussion of competing for people with high salaries requires business returns and cash to support it. Albert criticized treating a ByteDance-style approach—large numbers of top people and capital competing across parallel bets—as the default for ordinary teams. He noted that a sound business and culture can also help ordinary team members consistently do the right things. First determine whether the goal is personal income, a sustainable small company, or venture-backed expansion, then choose the approach.
Every project should retain at least four distinct categories of evidence: what users receive; what assets remain with the company after delivery; how money returns to the operating business; and why capital providers are willing to assign a valuation. Financing can improve supply and survival conditions, but cannot replace renewals. An exit may provide returns to some shareholders without demonstrating that the original product was profitable. A founder's technological ambition cannot replace the user's task. In particular, accounts of founder psychology and transaction motives in the Character.AI material must retain their attribution as Qu Kai's inferences, rather than turning hindsight into an inevitable destiny.
A minimal implementation for a product team is a one-page review each month: the most important current user problem; three controllable conditions; three external assumptions; the next empirical test; cash and time limits; who owns the result; and conditions for stopping. Team size and tool adoption are means. If user value, operating results, and personal goals conflict, write down the conflict and make a choice instead of covering it with a larger story.
2026-10-20 Addendum: Short-term new-word tool sites — how to turn hot trends into deliverable orders
Xu Xuan (@xuxuanzhou) published a post-mortem on October 19, 2026, about a Bulin 47 meme video tool site. The product does only one thing: users upload a photo and generate a finished video in which they appear in the meme video. Below is a methodological collation of this single case; business figures and timelines all come from the author’s self-report and attached images, have not been independently audited, and cannot be used to predict revenue for other sites.
First distinguish turnover, listed price, and profit
The author opens by describing cumulative transaction volume using the tax-inclusive Stripe gross transaction value, then clarifies in the "Data" subsection: the $5,480 in 72 hours referenced in the title refers to the listed order prices in the 72 hours after the first order, excluding checkout taxes and before deducting refunds; the first-week figures of 558 orders, 393 paying customers, and $6,434 also use the listed price metric, with 22 refunded orders. The attached figure additionally shows a cumulative gross transaction value of $7,197.19 and net sales of $6,062.52, which cover a different time range from the two preceding figures. The original text also notes that as of the time of posting, API expenses exceeded RMB 10,000. These amounts must not be mixed in calculation, and "net sales" must not be treated as profit after deducting model, payment, operations, and labor costs.
Find the “want to replicate” task, then judge the window
The author first tested a hotel lobby page on an old site, then had Claude Code help find similar new memes from TikTok and used Google Trends to compare candidate terms. His reason for choosing Bulin 47 was that the trend was just emerging, replication demand had appeared, and related domain names were still registrable. What is worth reusing here is connecting popular content to a specific tool task, rather than only chasing search volume. Trend curves, available domain names, and a small number of success cases are insufficient on their own to prove paid demand.
The author reports that the order peak was very short: from October 10 to 15, daily orders were 180, 185, 100, 33, 24, and 25 respectively, and he states that 85% of revenue was concentrated in the first 72 hours. This suggests that short-term tools should prepare login, payment, credits, and event tracking in advance to reduce the time from opportunity discovery to delivery; one cannot extrapolate this curve to the lifespan of all hot trends, nor use “speed” to skip basic acceptance checks for payment, generation, and download.
Pricing must match the usage task; orders appearing after a price change do not establish causation
The author found that the default recommended monthly subscription was unsuitable for users who “make one video and leave,” so he switched to prioritizing display of credit packs; his records show the first order appeared about eight minutes after the change, and he states that $9.99 single-video orders accounted for 92%. This supports a yet-to-be-validated judgment: for one-time consumption tasks, one can first compare the fit of pay-per-use or credit packs versus subscriptions. It is not a conclusion that all AI tools should abolish subscriptions, and the before-and-after comparison does not exclude the influence of traffic, timing, and other page changes.
Business rules written by AI must also be reviewed, not just whether the code runs
The main text and fault-record screenshots show that AI once added a daily limit and a rule that “if the person is not changed, judge it as failed,” causing normal paid finished videos to be rejected or automatically refunded. Other faults involved database connections, mobile uploads being obscured by navigation, inability to find the generation entry after payment, photos being lost before and after login, and upstream finished videos failing to be transferred and saved successfully. Editorial recommendation: before launch, confirm item by item who approves limits, what evidence counts as failure, and how refunds are triggered; walk through a real process along “payment—upload—generation—result saving—download.” Retain necessary security, copyright, and anti-fraud constraints; fixing misjudgments should not be misunderstood as deleting all protections.
Use segmented funnels to locate problems, and keep boundaries for conclusions
The author's attached chart lists the "completion rate after opening the payment page" as 54% for iPhone and 17% for Android, and separately lists the payment rate for all visitors by source as 5.8% for direct visits, 3.9% for ChatGPT, and 2.3% for Google. These use two different denominators and cannot be substituted for each other; payment methods, region, and user intent were also not controlled for. The author's explanation that Apple Pay is smoother remains a hypothesis, and the higher conversion rate of ChatGPT visitors does not prove that llms.txt or IndexNow drove growth. The original text also records that after sending a $6.99 discount email to 100 users who abandoned their carts, none made a purchase, which shows that a single price cut did not recover this batch of lost users, and cannot be generalized to mean that recall efforts are permanently ineffective.
When to use: When facing a tool demand that fades quickly and has a clear target action, first record the discovery time, demand evidence, delivery cost, and stop condition, then observe failures at each step. What is reusable is the validation and delivery process; profitability still needs to be judged by results after deducting refunds and all costs within the same cycle.
Source:Full Post-Mortem of Xuxuan's New Word Site (X, 2026-10-19). This article is an independent compilation; the original text and six accompanying images are retained in a private knowledge base.
October 20, 2026 addition: turn a short-lived search trend into an order you can deliver
On October 19, 2026, Xuxuan (@xuxuanzhou) published an account of a tool built around the Bulin 47 video meme. Its task was narrow: users uploaded a photo to generate a video placing themselves in the meme. This section extracts methods from that single case. Business figures and the timeline come from the author's account and attached images, have not been independently audited, and cannot predict revenue for another site.
Separate transaction volume, order prices, and profit
The opening uses Stripe transaction totals including tax to describe cumulative volume. The later data section clarifies that the headline $5,480 represents order prices during the 72 hours after the first order, excluding checkout tax and before refunds. The first-week figures of 558 orders, 393 paying customers, and $6,434 use the same price basis, with 22 refunded orders. A separate screenshot shows cumulative transaction volume of $7,197.19 and net sales of $6,062.52 over a different reporting window. The author also reports API spending exceeding RMB 10,000 by the time of publication. These figures must not be combined, and net sales are not profit after model, payment, operating, and labor costs.
Identify the task people want to reproduce, then assess the window
The author first tested a hotel lobby page on an existing site, then used Claude Code to help find similar emerging TikTok memes and compared candidate terms in Google Trends. Bulin 47 was selected because interest was beginning to rise, people wanted to reproduce the video, and relevant domains were still available. The useful method is connecting popular content to a specific tool task, rather than chasing search volume alone. A trend curve, an available domain, or a handful of successful examples cannot independently establish willingness to pay.
The reported order window was short: daily orders from October 10 through 15 were 180, 185, 100, 33, 24, and 25. The author states that 85% of revenue arrived in the first 72 hours. This suggests preparing login, payments, credits, and analytics before an opportunity appears to shorten the path to delivery. That one curve does not establish the lifespan of every trend, and speed does not justify skipping basic checks of payment, generation, and download.
Match charging to the task; orders after a pricing change do not establish causation
The author found that a default monthly subscription did not fit users who wanted to make one video and leave, so credit packs became the primary offer. The account places the first order about eight minutes after that change and says $9.99 single-video orders represented 92% of orders. This supports a hypothesis worth testing: for an occasional task, compare per-use pricing or credits with subscriptions. It does not mean every AI tool should remove subscriptions. The before-and-after observation also does not control for traffic, timing, or other page changes.
Review AI-generated business rules as well as working code
The text and incident log describe daily limits and a rule that marked a result as failed when it appeared not to replace the person. These rules rejected normal paid outputs or triggered refunds. Other incidents involved database connections, mobile navigation covering uploads, missing generation entry points after payment, photos lost across login, and outputs not being successfully copied from the upstream service. An editorial application is to confirm who approved each limit, what evidence constitutes failure, and how refunds are triggered. Test the complete payment–upload–generation–saved result–download journey. Necessary safety, copyright, and fraud controls must remain; fixing false positives does not mean removing all protections.
Use separate funnel stages and retain uncertainty
An attached chart reports payment completion after opening checkout at 54% for iPhone and 17% for Android. Another reports payment as a share of all visitors by source: 5.8% for direct visits, 3.9% for ChatGPT, and 2.3% for Google. These have different denominators and are not interchangeable. Payment methods, geography, and intent were not controlled. The proposed Apple Pay explanation remains a hypothesis, and stronger conversion from ChatGPT visitors does not establish that llms.txt or IndexNow caused growth. The author also reports that none of 100 checkout abandoners bought after receiving a $6.99 offer by email. That discount did not recover this group; it does not establish that recovery campaigns never work.
When to use this: for a clearly defined tool task with rapidly declining demand, record when the opportunity was discovered, evidence of demand, delivery costs, and stop conditions, then observe failures at each step. The reusable lesson is the validation and delivery process. Profitability still requires a consistent reporting period with refunds and all costs deducted.
Source: Xuxuan's full account of the emerging-keyword site (X, October 19, 2026). This is an independent synthesis; the original text and six images are retained in the private knowledge base.