OPC 的真正难题，是 AI 还没学会替你把东西卖出去
Summary
This Keji Luandun episode treats OPC as an AI-era One-Person Company narrative rather than only a legal registration form. The hosts argue that AI and Vibe Coding lower the cost of making pages, scripts, apps, content, and short-video workflows, but do not automatically solve product choice, Customer Pull, sales, payment collection, compliance, delivery, or industry judgment. Its practical advice is to find a real customer and a sellable offer before treating company registration, park subsidies, offshore entities, or “agent teams” as the source of value.
Key Claims
- The episode distinguishes legal one-person limited companies from the current AI slogan that one person can run all company functions with AI.
- One-person legal structures still require financial separation, tax handling, contracts, and responsibility boundaries; AI does not remove those obligations.
- Industrial parks, incubators, subsidies, token grants, or registered addresses can create startup activity signals without proving demand.
- AI coding makes the build step faster, but the harder questions are what to build, who wants it, how to sell it, how to collect money, and how to support customers.
- Successful OPC examples are framed as domain experts or people who already know how to sell; AI amplifies their execution rather than creating business judgment from nothing.
- AI short-drama and game examples show that generated assets or code still need directors, game producers, taste, pacing, and professional review.
- B2B use cases such as proposal writing remain difficult because format, procurement, context, trust, and adoption cycles are not solved by text generation alone.
- Overseas OPC narratives may offer higher willingness to pay, but foreign company registration, tax filings, accounts, fund movement, and compliance make the operating burden heavier.
- AI training, coaching, and “OPC陪跑” businesses can monetize the concept itself; that does not prove OPC as a general startup model.
- The hosts close on a sales-first rule: if a real client wants to pay and needs a contract, company setup can follow; registration before demand is mostly distraction.
- App and mini-program creation may grow because AI lowers supply cost, but platforms such as App Store, Apple, and WeChat may capture more reliable upside than individual low-demand apps.
Key Quotes
“AI 还没学会替你把东西卖出去” — the episode’s title and central commercial boundary.
“OPC 不重要，重要的是怎么用 AI” — the hosts’ summary stance against concept chasing.
“先找到第一个客户” — the practical registration-before-demand warning.
“最挣钱的 OPC 可能是教别人怎么做 OPC 的 OPC” — the episode’s caution about training-driven hype loops.
Connections
- Keji Luandun — show context for this AI, startup, and commercialization discussion.
- One-Person Company — new concept page for the episode’s OPC frame.
- AI Engineering Thinking, Vibe Coding, One-Shot AI Coding, and AI Coding Verification — AI can accelerate execution but still needs decomposition, tests, review, and responsibility.
- Customer Pull, Pre-Product Selling, Distribution Led Product Building, and Product Led Willingness To Pay — the demand and sales layer that the hosts put before company registration.
- AI Commercialization Pressure — personal-startup version of the broader AI business-closure problem.
- Human Judgment Under AI and Domain Expert Alignment — domain knowledge, taste, and final responsibility remain with the operator.
- AI Interactive Entertainment, Video Models, and AI Game Industrialization — content and game examples where generation helps but does not replace professional craft.
- AI Agent Overseas Commercialization, Cross-Border Fund Transfer Risk, and Capital Account Investment Restrictions — overseas market and account-compliance cautions.
- Software Payment Culture, App Store, Apple, and WeChat — platform and payment-context layer behind the question of who earns when AI creates more apps.
Contradictions
- No direct contradiction with prior wiki content. The episode reinforces existing warnings that faster AI building does not remove AI Engineering Thinking, Customer Pull, Product Led Willingness To Pay, or Human Judgment Under AI. Legal, tax, and overseas-account remarks are episode-level guidance and should not be treated as verified legal or tax instructions.