Podcast Atlas
entity Updated 2026-07-09 Tags: Podcast, Media

Keji Luandun

Keji Luandun is the podcast context for 付费片花：平台的暴力抵抗与互联网大厂的隐形税收, 我们把 AI 塞进花店后，才知道AI落地有多脏, 困在系统里的酒店，你不知道的携程垄断练成史, 阿里千问离职余震，在几万人的铁球里如何体面生存, 从QQ会员到豆包包月，中国人为什么总觉得软件该免费, AI 会写代码了，为什么你还是做不出产品？, 当“印钞机”百度开始失血，是天灾还是人祸？, 那个不穿西装的程序员，扯出了国产操作系统二十年秘史, 为什么Manus必须出海？聊聊国产大模型的“文科生困境”, OPC 的真正难题，是 AI 还没学会替你把东西卖出去, 把 AI 吹成核武器的人，亲手拉下了新冷战铁幕, 当华为抛出韬定律，我们该信它到哪一步？, 别在国内卷了，去美国看看只要产品好就有人付费的市场, 除了石油和海峡，这届伊朗战争开始算计你的服务器了, 这半年，我们又买了哪些科技好物？, 智力贬值的春节见闻录，与那场正在酝酿的优贷危机, 当可靠的代码变成了偶尔发疯的OpenClaw，我们未来的工作范式变迁, and 把身体数据存起来，可能是普通人最划算的 AI 投资. Across these sources, the show discusses Chinese AI companies, model strategy, product commercialization, organizational order, user behavior, platform shifts, software payment, practical AI-enabled work, offline retail AI, domestic operating-system history, online-travel platform governance, local-service merchant economics, AI-agent overseas commercialization, AI-era one-person-company narratives, AI export-control geopolitics, Huawei-style semiconductor strategy, overseas market selection, cloud infrastructure under wartime risk, everyday technology purchasing, AI’s pressure on labor value and credit assumptions, local-agent reliability and safety, personal health data, and AI-era education choices.

The Huawei episode adds a semiconductor and organization-strategy branch. It argues that Tau Law is more credible as a latency-oriented KPI and Constraint Driven Engineering Strategy than as a literal replacement for Moore’s Law, then connects Semiconductor 3D Stacking, HiSilicon, Ren Zhengfei, and Huawei Organizational Methodology to explain why a technical route can also become internal mobilization and public narrative.

The Ctrip episode adds a non-AI platform-governance branch. It uses Ctrip / Trip.com Group, Liang Jianzhang, Ji Qi, Shen Nanpeng, Fan Min, Wu Hai, Qunar, Elong, and Tongcheng Travel to explain how Chinese online travel concentrated around operational infrastructure, supplier systems, and capital integration. Its main conceptual contribution is the connection among Online Travel Agency, OTA Platform Concentration, Hotel PMS Inventory Control, Hotel Platform Pricing Power, Travel Booking Hidden Fees, Homestay Differentiation, Platform Antitrust, and Platform Data Regulation.

The flower-shop AI episode adds an offline implementation branch. It argues that AI adoption becomes legible only after entering the store workflow: flower materials, hands-busy workers, A4 order sheets, platform prompts, customer substitutions, paid traffic, response rules, and printer/OCR workarounds define where AI can help. Its main concepts are Offline AI Implementation, AI Visual Merchandising, Operational Data Capture, and Local-Life Platform Dependency.

The flower/cake paid teaser adds a platform-intermediation branch to that same local-services material. It argues that live rooms and traffic operators can promise national one-hour fulfillment, then hand the physical work to nearby shops, creating Platform Intermediation Tax on top of Local-Life Platform Dependency: the shop keeps the messy fulfillment work while traffic ownership captures much of the margin.

The AI coding episode adds a grounded operator perspective through Lao Gao, Zhang Le, and Wang Dafu: AI can write code and scripts, but useful products still require AI Engineering Thinking, domain know-how, tests, logs, review, and human communication.

The Baidu episode adds a legacy-platform perspective: Baidu’s search-ad decline, weak Wenxin product mindshare, and missed mobile-era pivots are treated as evidence that old traffic advantages can turn into Cash Cow Strategic Inertia once Open Web Traffic Decline changes where users begin.

The domestic operating-system episode adds a China software-infrastructure perspective. It uses Tongxin Software, Deepin, Hiweed Linux, Tongxin UOS, and Kylin OS to show how a community Linux lineage can become part of Xinchuang Operating Systems, and how customer structure can push an organization from developer culture toward sales, delivery, certification, and hierarchy.

The Manus episode adds an AI-agent market perspective. It treats Manus as a workflow product whose strongest use cases are overseas SEO, advertising, browser operation, and foreign-trade marketing, then connects its claimed Meta acquisition to AI Agent Overseas Commercialization, China Agent Market Friction, and Chinese Model Liberal Arts Constraint.

The OPC episode adds a personal-startup perspective. It argues that One-Person Company enthusiasm overreads AI’s ability to replace business functions: AI can help one person build faster, but the operator still needs Customer Pull, sales ability, legal and tax responsibility, delivery judgment, and a reason for customers to pay.

The AI export-control episode adds a policy and geopolitics perspective. It argues that treating frontier models as nuclear-level capabilities can create AI Safety Narrative Backfire, where governments respond through AI Export Controls, Frontier Model Access Restrictions, and an AI Cold War frame that weakens closed model companies’ SaaS-like reliability while making Open Source AI Models such as DeepSeek and GLM 5.2 more attractive.

The Win U.S. trip episode adds an overseas-market and field-observation branch. It uses Tesla Robotaxi, Waymo, CES, Bay Area startup gatherings, Chinese AI model usage in the U.S., and Manus’s direction change after a U.S. visit to argue for Payment Led Market Selection: founders should choose markets by real demand and payment behavior, not by a default Software Payment Culture or “China first” habit.

The data-center war-risk episode adds a physical infrastructure and geopolitics branch. It starts from updates on 你的书房 and 热乎乱炖, then argues that cloud regions, data centers, submarine cables, and AI compute facilities can become wartime pressure points. Its main contribution is the connection among Digital Infrastructure War Risk, Data Center Physical Resilience, War-Aware Disaster Recovery, Regional Network Topology Risk, Asymmetric Infrastructure Attack, and AI Compute Continuity.

The tech-purchase episode adds a consumer and small-team infrastructure branch. It uses Pinduoduo subsidies, Mac mini and iPad purchases, NAS drives, UPS protection, desktop chargers, small monitoring screens, OLED displays, net-disk plans, and travel gear to show Personal Infrastructure Cost Accounting in practice: gadgets are judged by solved workflow problems, ownership cost, backup reliability, and daily friction rather than by specs alone.

The intelligence-devaluation episode adds a social and credit-risk branch. It connects Seedance-style video generation, GLM5 coding, AI copyright, digital celebrities, flower-shop fieldwork, and AI company organization to Intelligence Devaluation and Prime Borrower Credit Risk: if AI makes cognitive skill cheaper, product builders need more domain knowledge, and lenders may need to question whether old white-collar borrower signals still imply stable repayment.

The OpenClaw reliability episode adds a local-agent safety branch. It argues that Open Claw is compelling because it can use local machines, accounts, browser state, memory, and AI Skills, but those same affordances make it Probabilistic Software rather than traditional reliable code. The episode extends the show’s AI-work synthesis into Agent Permission Boundaries, Local Agent Execution, Model Routing Cost Control, Model Context Protocol, and Human Judgment Under AI by showing why users should isolate devices, narrow permissions, verify outputs, and avoid giving agents direct control over money or key accounts.

The health-data episode adds a healthcare and education branch through Jiang Xun / 江迅. It argues that Personal Health Data may be one of the highest-value ordinary-person AI investments because long-term reports, wearable signals, and Continuous Glucose Monitoring curves can give AI Health Management enough context to flag trends before a single hospital visit crosses a threshold. The episode keeps Human Judgment Under AI central by insisting that AI should assist health management and doctors, not replace diagnosis or treatment. Its second half extends the show’s AI-era personal strategy through The Fifth Dimension / 第五维度, College Major Choice, Learning How To Learn, and Distribution-Out Personal Strategy.

The Asahi ransomware crossover with Top of Japan adds a cybersecurity and business-continuity branch. It uses Asahi Group / 朝日集团, Super Dry, Qilin Ransomware Group, SAP, and Sony Pictures to show how cyberattacks move from encrypted files into factory stoppage, supermarket shortages, delayed reporting, data-publication risk, and ransom decisions. Its main conceptual contribution is the link among Ransomware Business Continuity, Offline Backup Recovery Drills, and Personal Security Tiering: backups, recovery drills, and risk-based account discipline are treated as the practical basis for refusing extortion and reducing personal attack payoff.

Connections