副业日报

· AI 驱动生成

信号

Personal finance tools remain consistently popular on HN, with clear willingness to pay in decision-scenario contexts. The author's use of personal pain points to drive the product indicates a real demand unmet by major tech companies.

为什么现在还能做

In 2026, personal financial scenarios have become more complex (salary, mortgages, investments), making single calculators insufficient. Evidence shows users are willing to pay for 'multi-scenario comparison + tax calculation' ($69/year).

建议怎么做

1) Start with your own financial management pain points and build an MVP using Next.js + OpenAPI. 2) Introduce a multi-scenario comparison page. 3) Launch via Hacker News and other channels, then iterate on the differences between free and paid versions based on feedback. 4) Consider expanding to overseas markets (English-first, with multi-currency support added later).

变现说明

Free version opens the core engine; paid version is $9/month or $69/year, unlocking multi-scenario comparisons and Monte Carlo simulations. Early user discounts lock in pricing until cancellation, improving renewal retention.

可复制性

Medium barrier. Relies on the accuracy of tax logic (region-specific), but the framework logic is reusable; competition comes from similar calculators, with differentiation lying in UX prioritizing scenario comparison.

机会视角：Validate one specific scenario first before discussing multi-scenario comparisons; paid validation is more important than feature stacking.

信号

Shifting AI agent workloads from cloud to local devices represents a new direction under dual pressure of token costs and privacy concerns in the industry.

为什么现在还能做

In 2026, agent usage frequency has risen, making cloud billing models unsustainable. Local inference technology is maturing, and hardware thresholds have dropped to consumer-grade RTX cards.

建议怎么做

1) Pay attention to the local deployment method of Perplexity Portable Computer. 2) Test using DGX Spark or RTX 3090+ to understand model performance and token savings. 3) Explore how to migrate existing cloud services to local setups to reduce costs and enhance privacy protection.

变现说明

Revenue through selling hardware like DGX Spark; for users, local processing saves cloud token fees.

可复制性

High replicability. Open-source models and standard toolchains can be mimicked, but strong engineering integration skills are required.

机会视角：If you're building AI agents, consider local deployment to reduce costs and protect privacy; this is a future trend.

信号

Acquires users through unlimited free core features (background removal) while promoting premium features (Nano Banana image generation) via an points system—a typical 'freemium + gamification' growth model.

为什么现在还能做

Content creators' demand for efficient image processing continues to grow, and they are willing to pay for batch/high-quality generation. Vibe Coding emphasizes rapid prototype validation, lowering development barriers.

建议怎么做

1) Use a Vibe Coding platform to quickly build an MVP for a similar AI tool. 2) Design unlimited free core features (e.g., background removal) to attract users. 3) Distribute points through check-ins, registrations, etc., guiding users to try paid features (e.g., large-batch image generation). 4) Collect user feedback to iterate on the product.

变现说明

Main revenue comes from premium feature subscriptions (e.g., Nano Banana bulk image generation) or user conversion driven by the points system.

可复制性

Low technical barrier; core lies in product design and growth strategy. Highly replicable, but requires continuous optimization of the balance between user experience and monetization.

机会视角：Free features are the hook; the points system is the lever. Don't fear giving away free features; fear that free won't drive paid conversions.

信号

A STEM-background individual with no app experience completed and open-sourced an independent Android app in one week using Claude Code + gpt5.4/GLM5.1, validating that vibe coding is feasible.

为什么现在还能做

The barrier for LLM APIs has dropped to the point where a single person can handle architecture + code + polish. Domestic models like GLM can also stably deliver, making it suitable for non-CS majors starting a side hustle.

建议怎么做

Select a刚需 small scenario (mistake notebook/note/tool), use Claude Code or OpenCode, install superpowers plugin for brainstorming and SOP constraints. Use top-tier models first for architecture and core code (~75%), then switch to cost-effective domestic models for polishing, and deploy to GitHub for traffic.

变现说明

Tool apps can go free + in-app purchases/subscriptions or one-time purchase. Initially rely on word-of-mouth and GitHub stars for organic downloads.

可复制性

Low barrier: depends on model capability rather than scarce resources. The threshold lies in topic selection and execution speed; highly replicable.

机会视角：This is doable. Don't wait for perfect technology; run the MVP first. Single model + SOP is more stable than chasing multi-agent setups.

信号

A vertical exam prep product combining official IELTS practice tests + AI speaking mock exams + AI writing feedback. An indie developer built the full stack, with features tightly aligned to user pain points.

为什么现在还能做

Study abroad/exam categories have stable willingness to pay. AI can handle high-frequency, low-cost training (speaking/writing), while human teachers only need to provide diagnosis, leaving room for asymmetric competition.

建议怎么做

Lock onto specific exam pain points (listening intensive practice, speaking shadowing, essay detailed feedback). Build full-stack with Next.js/React, deploy on Cloudflare Workers. Use Clerk for accounts, Stripe for payments, and integrate Workers AI + Azure Speech for writing/speaking. Offer basic content for free, unlock core evaluations with PRO subscription.

变现说明

PRO membership at 79 CNY/month or 499 CNY/year unlocks full intensive listening, AI writing scoring, AI speaking scoring, and high-score templates. Free practice tests drive traffic.

可复制性

Medium barrier: vertical content copyright and evaluation quality are key. Requires official test authorization or self-built question banks, plus a stable AI evaluation pipeline.

机会视角：The track is viable, but don't package it as a score-boosting magic pill. Build a tool users open daily for training. Subscription pricing should match the user's exam preparation cycle.

信号

iOS/macOS photo map app with local processing, no account required, no uploads. Features heatmaps, posters, and annual summaries, aligning with privacy and tool-centric trends.

为什么现在还能做

Users are willing to pay for local-first, one-time purchase tools, especially among travel and memory-logging demographics with higher conversion rates.

建议怎么做

To replicate this direction: focus on local indexing of iCloud/photo libraries, providing heatmaps, filtering, and shareable posters. Technically, address location permissions and local image processing performance. Monetize via a freemium model with a one-time unlock for premium features.

变现说明

Basic features are free; premium features are unlocked with a one-time payment of ¥22. Existing users can unlock all premium features for free, relying on word-of-mouth marketing.

可复制性

Moderate barrier: Apple ecosystem photo APIs and location data processing have entry hurdles, but the core logic is generalizable. Differentiation lies in UI and privacy commitments.

机会视角：There is still a market for niche, high-quality one-time purchase tools. The key is to nail privacy and user experience; the ¥22 price point is an impulse-buy friendly range.

信号

Ramp's self-developed Inspect platform features remote sandboxes, internal data integration, and automated validation, handling 1 million sessions, 75% merged PRs, and 200+ upper-layer agents.

为什么现在还能做

General AI coding tools lack internal context and validation loops. Self-built platforms significantly outperform external solutions in complex engineering environments, validating the ROI of building custom tools.

建议怎么做

Reference path: use open-source coding harnesses (e.g., OpenCode) as a foundation, add remote sandboxes and internal MCP/API integrations, enforce backend testing and frontend screenshot validation. Platformize agents for debug, review, and oncall tasks.

变现说明

Indirect monetization: improved engineering efficiency and reduced manual repetitive labor. Can be exported as platformized services or consulting externally, but the primary value is internal cost reduction and efficiency gains.

可复制性

High barrier: requires engineering depth and integration with internal systems. However, the architectural approach is replicable for teams with internal data and processes.

机会视角：General tools can only handle 80% of the work; the remaining 20% is the moat. Building internally is not about showing off—it's about filling gaps in context and validation.

信号

MIT/BCG/S&P data reveals that most AI projects die from process flaws rather than model quality, offering an executable 5% success paradigm.

为什么现在还能做

SME AI adoption surged from 40% to 58% in a year, and companies using AI have 2.3x higher probability of revenue growth. However, most remain stuck at the 'buy tool, run pilot' stage.

建议怎么做

Define constraints before capabilities: set price floors, strategy whitelists, and exception escalation for agents. Reintroduce humans into workflows to handle exceptions. Assign an Owner for every output and establish an audit trail.

变现说明

For service providers/consultants, this is sellable AI implementation consulting and outsourced operations. For enterprises, it avoids failure losses and provides access to the 91% of organizations achieving actual revenue growth.

可复制性

Low barrier to methodology, high barrier to execution discipline. The key is organizational habits, not technical selection.

机会视角：Stop treating AI as magic. Clarify accountability chains and human-machine division of labor first. Deployments without an Owner are just passing the buck.

信号

An indie developer successfully runs a complete payment, customer service, and SEO loop from 0 to 1, targeting a specific刚需 niche (classroom/meeting simultaneous translation).

为什么现在还能做

AI capability下沉 makes it possible for non-large teams to build stable real-time speech-to-text tools, but good interaction design and scenario adaptation remain barriers.

建议怎么做

Identify a specific audio scenario (e.g., live broadcast subtitles, meeting records), use existing APIs (Whisper/Deepseek) to wrap a simple web tool. Focus on reconnect resilience, latency optimization, and export features (SRT/TXT). Integrate Stripe for payments.

变现说明

Monetize through paid subscriptions or one-time unlocks (e.g., long-record export, AI summaries), covering API costs with profit margins.

可复制性

Medium technical barrier. Main moats lie in scenario understanding, UX refinement, and continuous iteration/maintenance. Suitable for developers with frontend experience to replicate.

机会视角：This validates that indie development is not just about writing code—it's about running a complete business. Even with one deep feature, you can survive.

信号

Low-price, high-performance hardware ($90 8-inch tablet) appears at best price of the year on Amazon channels, suitable for DIY side-business scenarios like home monitoring, digital frames, car entertainment, or low-cost content distribution terminals.

为什么现在还能做

Hardware prices have hit bottom, reducing trial-and-error costs for hardware modification and content deployment. Although Fire OS is closed, sideloading or web-based usage is possible.

建议怎么做

Monitor promotional nodes on Amazon Global Store or domestic channels, buy in bulk, and modify into specific scenario devices (e.g., kindergarten monitoring, elderly care screens, retro game consoles). Or use as low-cost test devices for multi-platform app adaptation.

变现说明

Hardware price arbitrage (tariffs and logistics require caution), premium resale after DIY modification, or low-cost carriers for content distribution.

可复制性

Information asymmetry arbitrage; requires monitoring promotional nodes and channel stability. Modification services need hands-on technical skills.

机会视角：Low-priced hardware is often a side-business 'ammunition cache'. A $90 tablet can do many things; the key is finding a small need it can solve that others find too troublesome.

信号

Multiple countries treat "termination of tax residency" as a taxable event, triggering capital gains even without selling; meanwhile, 76 jurisdictions have committed to automatic CARF data exchange, narrowing the window for identity planning.

为什么现在还能做

Cross-border data exchange begins in 2027; Canada/Australia use exit pricing, the UK has a 5-year re-entry recovery mechanism, and the PR 0% window only applies to new cases applied for before 2027.

建议怎么做

Verify your tax residency and exit tax clauses with your holding location → Record cost basis and current fair market value → If expecting significant appreciation, complete residency switching before the rise and retain substantive residence evidence → Choose stable jurisdictions with clear legislative cycles rather than relying on verbal zero-tax rates.

变现说明

High-net-worth individuals have a strong willingness to pay for compliant planning services; this can serve as a paid entry point for cross-border tax/asset allocation consulting or as an execution service for family offices.

可复制性

Rules are public and paths are standardized; true scarcity lies in professional expertise regarding tax law details and residence fact-finding; information asymmetry will narrow as CARF expands.

机会视角：This calculation depends not on when you sell, but when you "leave"—positioning early ahead of market movements is far more cost-effective than saving taxes事后.

信号

AI data centers are hogging consumer-grade DRAM capacity; Q1 contract prices rose 93%-98% quarter-over-quarter, with storage rising from 16% to 23%-35% of PC BOM cost, hitting entry-level machines hardest.

为什么现在还能做

Industry predicts the sub-$500 entry PC segment may disappear by 2028; the supply-side structural price hike is not a short-term fluctuation but the start of a new cost floor.

建议怎么做

Monitor alternatives along the price hike chain: cloud desktops/remote hosts to reduce local hardware investment; offer low-spec terminals plus cloud computing bundled services to enterprises/education; avoid hard-loss pure hardware distribution and shift toward software-hardware integrated deployment and operations monetization.

变现说明

With narrowed hardware margins, profit comes from deployment, hosting, operations, and customized integration fees; B2B annual service fees are more cycle-resistant than one-time machine sales.

可复制性

The barrier is not price competition but service capability; IT delivery ability and industry know-how are core moats.

机会视角：The low-end hardware dividend period is over; the era of making money selling shells is turning a page. Only those who can bundle computing power and services will survive.

信号

The UK AISI has a new head and significant challenges. Nvidia spends $6 billion to 'reverse aquihire' Poolside. Hugging Face reportedly looks to sell for $13 billion. Use of Anthropic’s top model lags. Why Americans use chatbots for health information. And what role should AI play in schools?

为什么现在还能做

Based on signal consolidation from this edition

建议怎么做

Recommended to verify against the original text and conduct small-scale tests

变现说明

Monetization possible through productized services, subscriptions, or consulting

可复制性

Further replication feasible once clear demand is identified

机会视角：Opportunities for replicable products or services can be explored around this signal

信号

(Work in progress — same as the project it's about. More on that below) Almost everything written about React Native security is about protecting your users : the shipped app, the phone it runs on. Almost nothing is written about protecting you , the React Native developer.

为什么现在还能做

Based on signal consolidation from this edition

建议怎么做

Recommended to verify against the original text and conduct small-scale tests

变现说明

Monetization possible through productized services, subscriptions, or consulting

可复制性

Further replication feasible once clear demand is identified

机会视角：Opportunities for replicable products or services can be explored around this signal

信号

A CISO who sees a low CVE count and deprioritizes prompt injection is reading the scoreboard wrong. Prompt injection has held the No. 1 spot on the OWASP Top 10 for LLM Applications for three consecutive years . When two leaders of that list checked it against 6,639 labeled real-world incidents, it came back at No. 12. The drop measures visibility rather than danger, because the attack operates where a vulnerability scanner cannot see it.

为什么现在还能做

Based on signal consolidation from this edition

建议怎么做

Recommended to verify against the original text and conduct small-scale tests

变现说明

Monetization possible through productized services, subscriptions, or consulting

可复制性

Further replication feasible once clear demand is identified

机会视角：Opportunities for replicable products or services can be explored around this signal

信号

On a day of high-priced product announcements, Apple offered a crumb of comfort to cash-strapped customers: the Polishing Cloth dropped in price by 53%, from $19 to $9.

为什么现在还能做

Based on this period's screening signal

建议怎么做

Suggest verifying with the original text and testing in small steps

变现说明

Can be monetized through productized services, subscriptions, or consulting

可复制性

Can be further replicated once there is a clear demand

机会视角：Explore replicable products or services around this signal

信号

Job posting for a remote AI test engineer focusing on web, app, API testing, and building AI testing workflows from scratch.

为什么现在还能做

Based on this period's screening signal

建议怎么做

Suggest verifying with the original text and testing in small steps

变现说明

Can be monetized through productized services, subscriptions, or consulting

可复制性

Can be further replicated once there is a clear demand

机会视角：Explore replicable products or services around this signal

信号

Per-seat pricing is dying due to price hikes using up headroom, CIOs cutting traditional software to fund AI tokens, and the unit being unglued from value. Three new models are emerging: Consumption, Outcome, and Resolution.

为什么现在还能做

Based on this period's screening signal

建议怎么做

Suggest verifying with the original text and testing in small steps

变现说明

Can be monetized through productized services, subscriptions, or consulting

可复制性

Can be further replicated once there is a clear demand

机会视角：Explore replicable products or services around this signal

信号

PromoGoat tested if digital PR could improve AI visibility. Claude initially described their AI marketing platform as 'a generic project-management tool.' They audited AI responses across four engines and found 38% share-of-answer, but zero presence in category prompts and citation pool.

为什么现在还能做

Based on this period's screening signal

建议怎么做

Suggest verifying with the original text and testing in small steps

变现说明

Can be monetized through productized services, subscriptions, or consulting

可复制性

Can be further replicated once there is a clear demand

机会视角：Explore replicable products or services around this signal

信号

Rumors and leaks about Apple's first foldable iPhone, expected to be called iPhone Ultra or iPhone Fold, with specs like a 7.8-inch inner display, Touch ID, dual rear cameras, and a price around $2,399.

为什么现在还能做

Based on this period's screening signal

建议怎么做

Suggest verifying with the original text and testing in small steps

变现说明

Can be monetized through productized services, subscriptions, or consulting

可复制性

Can be further replicated once there is a clear demand

机会视角：Explore replicable products or services around this signal

信号

Image credit: Unsplash+ Global Science Back-to-School Special Event—Hurry to the Livestream for Benefits! On Friday, September 4 at 20:00, tune into the Global Science WeChat video account livestream to unlock back-to-school exclusive offers: Buy three copies of Global Science magazine for just 9.9 yuan with a flash sale! Super-value annual subscription discounts come with free magazines, merch, and English improvement goodies! Endless giveaways including special issues, merchandise, and magazines! Plus, half-price slots are up for grabs! This event is exclusive to the livestream, so click the [Reserve] button now! Written by | Wang Yu, Reviewed by | Dong Yuan Can you whistle? If so, try whistling in front of a mirror, and you’ll notice that the corners of your mouth pull back slightly, leaving a small circular opening between your lips. When you whistle, your lips and tongue form precise shapes that create two exact openings. Air flowing through these points generates a tiny, self-sustaining feedback loop. The tongue creates minute disturbances in the airflow, which propagate forward to the lips; these disturbances then bounce back as pressure waves into the oral cavity, interacting with the tongue again. This process happens hundreds of times per second, effectively shuttling sound back and forth until it locks into a clear, stable note. Image credit: Unsplash This operates on the same basic physics as blowing across the mouth of a glass bottle, where the air cavity inside produces sound. If there’s water in the bottle, the smaller cavity yields a higher pitch; with no water, the larger cavity produces a lower tone. Scientists call this phenomenon a Helmholtz resonator. In essence, when you whistle, your mouth acts as an excellent resonator, and skilled whistlers can change pitch by controlling the cavity shape with their tongues. Whistling works differently from normal speech. During speech, vocal cords in the throat close to leave a narrow slit and vibrate to produce sound. When whistling, however, the vocal cords remain fully open and silent. Whistling Is a Childhood Skill Yet some people truly cannot whistle. Francisco Ruiz, a mechanical engineering professor at Illinois Institute of Technology in the US, notes that for most people, inability to whistle is simply a matter of practice. However, a small fraction of the population is anatomically incapable of whistling. Ruiz, a senior professor specializing in engine flow and injection with over sixty years of experience, took up whistling as a low-cost, down-to-earth side project late in his career. At the Acoustical Society of America, he shared his research findings: measuring the airflow inside the mouth while whistling revealed that “you must maintain the correct lip geometry and the correct geometry between the tongue and the palate. People born with cleft palates—a congenital condition where the roof of the mouth fails to close completely—cannot whistle because they cannot form that narrow channel.” Cleft palate sufferers are rare. For most, whistling is more like a skill best learned early. Maureen Stone, a speech scientist at the University of Maryland School of Dentistry, compares adults learning to whistle to learning a new language later in life: “If you take an adult to a new country to learn a new language, they’ll likely always speak with an accent; but if you take a child, it’s not a problem.” Stone identifies puberty as the critical turning point. Whistling depends on precise control of lip and tongue muscles, whose plasticity diminishes once puberty begins. Training Wheels for Whistling Humans aren’t the only whistlers; walruses can whistle too. Their oral anatomy is better suited for it, but for most humans, the “sweet spot” for whistling is extremely small. If you didn’t learn before puberty, it becomes very difficult as an adult. To help adults who still can’t whistle find the right feel, Ruiz invented Flutino, a small 3D-printed device that guides lips into the precise geometric shape needed for whistling. With Flutino, Francisco Ruiz can whistle while playing the ukulele. Image credit: Francisco Ruiz “This device replaces your lips,” says Ruiz. “Place it between your lips, blow, and it shapes your mouth for you.” Uniquely, unlike other instruments, it works on both inhalation and exhalation. Ruiz conceived the idea during the pandemic, iterating through multiple prototypes before finalizing the design. Tests show the device can produce whistles up to 100 decibels. Ruiz is now trialing Flutino as a teaching tool. The Whistle Club at UCLA is testing whether people who can’t whistle can learn with it. One tester reported: “I never knew how to whistle before, but I could do it in under ten minutes using Flutino.” Perhaps this little 3D-printed gadget can indeed help more people learn to whistle. References: https://pubs.aip.org/asa/jasa/article/158/4_Supplement/A387/3390263/Experiments-on-the-flow-acoustics-of-human https://www.popsci.com/science/why-people-cant-whistle/ https://flutino.weebly.com/learn-to-whistle.html https://doi.org/10.1152/japplphysiol.00902.2016 Originally published via the WeChat Official Account “Global Science.” For reprints, reply “Reprint” in the account’s background message, or contact newmedia@huanqiukexue.com via email. Content may not be used for marketing or promotional purposes. - E-commerce Ad - Global Science August 2026 New Issue Now on Sale Click the image or read the full article Buy Now Click [Like] to receive our content updates promptly Read the original article Platform URL: http://www.jintiankansha.me/t/FdNRCeElKx

为什么现在还能做

Based on this period’s screening signals

建议怎么做

Recommend verifying with the original source and running small-scale tests

变现说明

Monetizable via productized services, subscriptions, or consulting

可复制性

Further replication feasible once demand is clearly established

机会视角：Opportunity to explore replicable products or services around this signal

信号

Hey folks, I see so many posts ‘shipping by the beach’ or when out with the kids on the weekend. Umm, good for you? I’m not sure it’s the flex people think it is. Remember when we thought AI would take all our jobs so we didn’t have to work hard anymore? Yeah, turns out that’s not a thing. Everyone’s just even more mad for making things. Myself included. But I do draw the line at 1. talking to agents every second I’m not at my laptop and 2. if I’m with the kids. Whether that’s just the mental context switching is too hard or I’m just not a very hard-worker. You decide 🙂 But that’s all not to say that I absolutely want agent access on my mobile for whenever the moment strikes. Unfortunately for me, it’s usually when I should be going to sleep. I end up just laying there doing the luge while all the ideas swirl around my head when I should be reading + sleeping. Ben’s Bites is brought to you by Gravitee Every AI agent needs an identity. Every action needs accountability. AI agents are becoming teammates. They should also be held to the same standard. Gravitee gives every AI agent a verified identity, enforces what they can access & records exactly what they did, so agent accountability is built in. Headlines Updates for Remote Control in Claude Code - you can now start a new Claude Code session from your phone, dropped connections between your laptop and phone recover themselves, and the sessions load faster on iOS. Claude Academy - Anthropic’s collection of courses, tutorials and job-specific guides to use Claude. All the resources are free and open to anyone. Also read: Anthropic’s approach to teaching and learning AI. You can connect ChatGPT to Apple Messages. It can search messages, catch you up on conversations, and draft or send replies from your Mac. It’s available in ChatGPT Work and Codex on desktop. Also from OpenAI: GPT-Image-2 can now generate transparent backgrounds in the API. GPT-5.6-Sol is now 20% cheaper in the API - $4/$24 for 1M input/output tokens. Grok Bot is now available to more users - all SuperGrok Plus, Cursor Pro+ and Cursor Teams subscribers now have access. Plus there’s a free trial for a week that’s enough to get you hooked. Claude Mythos 5 will power Claude Security for enterprise customers. This is the first rollout of Mythos outside Project Glasswing. Deepgram just dropped Flux TTS, text-to-speech built for live conversation. It responds in as low as 80ms, carries context across turns, and handles interruptions natively. Spin up a stream and test it on your stack.* My feed Skydive - an always-on agent across Slack, email, iMessage and CLI, with your choice of model. I’m an investor. (read more) Nvidia and SpaceX are planning to launch GPUs to space. A Mac app that turns your agent traces into memory. A skill to find inconsistencies, bugs and design errors in your product. One of the best anti-slop skills and 10 more to explore. Two “explain it simply” skills: /eli5 and /eli5-for-grownups. Stripe MCP - Analyse your Stripe data with agents. Shopify’s founder built an open-source git store. (inspired by Cursor) Andrew Ng’s list of skills you need to build and deploy AI applications. A museum of AI interfaces. A visual guide to AI chip architectures. Designing Grok Bot with Grok Bot. Is your website ready for agents? What if everything goes right for AI? Learnings from the Aluminium trade. Should you be in the business of selling data to frontier labs? deblaot.dev - open-source replacements for bloated vendor software. Spline rebuilt its 3D editor with Agent Mode and an MCP. FigmaTrace - computer-use dataset for design. Search for agents - Firecrawl’s developer index, Exa’s plugin for codex/Claude Code, and Parallel’s fast mode for 10x cheaper search. Sam Altman on David Senra’s podcast + transcript broken into small chapters. Subscribe now Afters Find me on X, Linkedin, or YouTube Read about me and Ben’s Bites 📷 thumbnail via @keshavatearth *sponsors who make this newsletter possible 🙂 Wanna partner with us for the next quarter? Email us at shanice@bensbites.com or k@bensbites.com Ben's Bites is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

为什么现在还能做

Based on this period’s screening signals

建议怎么做

Recommend verifying with the original source and running small-scale tests

变现说明

Monetizable via productized services, subscriptions, or consulting

可复制性

Further replication feasible once demand is clearly established

机会视角：Opportunity to explore replicable products or services around this signal

信号

IT Home, August 25 — The WeChat Official Account “Douyin Blackboard” today published the “Douyin Announcement on Governing Malicious Marketing Accounts Hype Regarding Corporate Information,” stating that the platform has recently detected malicious marketing accounts engaging in black- and gray-market activities such as fabrication, insult, denigration, and commercial defamation, seriously harming enterprises and users’ legitimate rights and disrupting platform ecology. IT Home notes that “Douyin Blackboard” also publicized several cases, including one in which a well-known tech brand was long targeted by 59 accounts that aggregated and centrally published negative information spanning product incidents, stock price movements, and public controversies. Details: Case 1: Misleading the public by fabricating false corporate information Behavior includes inventing false claims of corporate illegality, quality fraud, or financial turmoil, or taking素材 out of context to mislead the public. For example, 13 marketing accounts edited images and twisted backgrounds to fabricate unfounded claims that a famous sports brand “pampers some while trampling others”; relevant accounts have been penalized. Additionally, 236 accounts fabricated false information about a well-known domestic beverage brand, tying it to nationalistic sentiment for marketing monetization; 86 severely offending accounts were punished, while others were guided through health-score mechanisms. Case 2: Water-arm gang attacks on fixed brands via matrix accounts Matrix accounts post in coordination with identical talking points, flooding platforms with batch submissions to steer sentiment against a company. For instance, 22 accounts periodically and集中ly published templated hit pieces against a famous insect-repellent brand, using news-style formats to cite questionable negative information as “avoidance guides.” Submissions were highly homogenized, video structures heavily templated, and posting followed a weekly schedule; 16 accounts were penalized, with remaining accounts guided via health scores. Case 3: Long-term aggregation of corporate negativity for malicious traffic diversion Behavior involves continuously collecting scattered negative information about a company and deliberately producing negative compilation videos to damage reputation. For example, 59 accounts long aggregated and centrally published negative content about a famous tech brand—including product incident scenes, stock fluctuations, and public controversies—while displaying extreme, attack-oriented, and disparaging on-screen copy urging users not to purchase the brand. Among them, 28 accounts were identified and penalized by the marketing-account model; others were guided through health scores. According to “Douyin Blackboard,” over the past month, the platform has, under its “Account Health Score Rules,” handled 1.4534 million违规 short videos and 597,500违规 accounts related to malicious marketing, including measures such as reduced recommendations, takedowns, and restrictions or suspensions of posting and monetization features.

为什么现在还能做

Based on this period’s screening signals

建议怎么做

Recommend verifying with the original source and running small-scale tests

变现说明

Monetizable via productized services, subscriptions, or consulting

可复制性

Further replication feasible once demand is clearly established

机会视角：Opportunity to explore replicable products or services around this signal

信号

Can you play PC games with the Logitech MX Master 4 mouse? Yes. Yes, you can. Alrighty, thanks for reading this week’s Boost Mode newsletter, see y’all next Tuesday. Okay, I realize that’s a bit facetious. You know you can play PC games with any old mouse, and I promise that I’m not wasting your time here. A better question might be, should you play PC games with the Logitech MX Master 4, the current version of the ultimate productivity mouse? And it’s more relevant than you might think. A lot of people build an elaborate desktop PC for both work and gaming. And if they’re anything like me, they prefer to keep their inputs consistent—that is, use the same stuff for work and play. To say nothing of the fact that this thing doesn’t come cheap at $120. Welcome to the Boost Mode newsletter. Every week I share my insights on PC gaming, desktop setups, and hardware optimization. I also track down some of the most interesting PC and nerd news from around the web, plus a PC game pick you might not have seen elsewhere. Want this newsletter to come directly to your inbox? Sign up on our website! Logitech’s MX Master is the pinnacle of productivity mice That’s easy enough for keyboards and headsets and monitors, but finding the perfect mouse for multiple applications remains frustratingly hard to come by. If anything, I’d say that gaming mice like the newly unveiled Razer Naga Pro V3 are better suited to this. Gaming mice are better at work than work mice are at games. They’re more flexible and versatile, and have higher performance. The only real downside is that if you’re using a wireless mouse for hours of gaming a week, you’ll need to recharge it more often. Eugen Wegmann But if there’s any “regular” square peg mouse that someone might be tempted to shove into a round multi-function hole, it’s the MX Master. Logitech’s flagship is generally agreed to be the king of standard PC mice, and has remained so since its initial model succeeded the MX Revolution over a decade ago. This mouse is so beloved that it’s often spotted on high-end Mac setups. And if Apple fans are willing to ditch first-party hardware, you know it’s good. I’ve used MX mice before, but for various reasons tend to go off them, for the aforesaid Razer Naga or a cheaper alternative. The MX Master 4 caught my attention, because it adds one more button…which, combined with that distinctive side wheel, brings it up to my required seven extra buttons in addition to the primary left, right, and scroll wheel. Why do you need seven extra buttons, you weirdo? Wait, what? I need seven extra buttons on a mouse? Yes. Here’s how it goes. I use my mouse as my primary media management tool, plus extra buttons for gaming. So I like to have buttons for volume up and down, play/pause, and mute. That’s in addition to gaming buttons, and I prefer three of them. In a typical action game that’s melee, throwable or special attack, and one more—I often bind that one to a map, for example. So that one extra button on the MX Master 4 is a big deal for me, specifically. It allows me to set those three long buttons for game functions, the haptic button on the base for mute, the button above the primary wheel as play/pause, and the side wheel as volume adjustment. Logitech This is weird, yes. It’s also kind of ignoring the action wheel, Logitech’s big push to expand the MX Mouse’s functionality more or less indefinitely. Anyway, that’s how I use mice—I use equivalent layouts on the Razer Naga (six buttons on the side) and Solakaka (three on the side, plus four more on top). I spotted a refurbished deal for the MX Master 4 bringing it under $90, direct from Logitech, and snatched it up. I wanted to see if I could well and truly have one mouse to rule them all. The Logitech MX Master 4’s hardware So how does the MX Master 4 measure up? In gaming terms, it’s a little weak, to be honest. Its “Darkfield” sensor is more interested in tracking on any surface you might throw at it, with a maximum DPI of 8000. That would have been mind-blowing a decade or two ago, and it’s eight times the sensitivity of the original MX Master. But compared to the $100 Razer Viper V3 Pro SE, it has less than one-quarter the sensitivity. Logitech doesn’t say what the polling rate is for the MX Master 4, because most people who buy one don’t care about it. It does have an option for 2.4GHz wireless or Bluetooth (and you absolutely don’t want to use Bluetooth), which won’t show any noticeable lag for most mere mortals. It doesn’t work over a wired USB connection. For those of you who really like to dig into the minutiae, it is 128.15mm long, 88.35 mm wide (at its widest point, that long spur for your thumb), and 50.8 mm high. The weight is 5.29oz, 150 grams. And if that last one made you wince, yeah, imma get to it. The Logitech MX Master 4’s software Software is crucial for gamers…at least the gamers that demand customization. One of the things that makes a keyboard a “gaming” keyboard, by my definition, is the option to create a new binding and setting preset that automatically launches with a game. Logitech’s G series mice offer this. The MX Master 4, managed by the Options+ program, does this too…technically. You can add a separate program right there in the menu, but it doesn’t seem particularly good at auto-detecting games. I was able to add them manually by digging into game files, and successfully bound Q and E to the thumb buttons in Baldur’s Gate 3 to move the camera around. There’s one hurdle passed. Unfortunately another core element of gaming mouse software is macros, combinations of key presses for complex inputs. Options+ does let you do this, but with some big limitations. You can do pretty much any combination of Ctrl, Alt, Shift, and Win buttons with alphanumeric keys and…that’s it. More complex combos and custom timings are out. But frankly all of these features are things I rarely use. For those three dedicated gaming buttons, I generally set them to the numerals 1, 2, and 3, and then set up singular binds in the game as needed. And Logi Options+ handles that okay. With the program bindings, it’s better than I was expecting. Here’s the annoying part: Unlike many gaming mice, the MX Master 4 does not retain bindings in its own internal memory. So the 1, 2, 3 preset didn’t stick when I brought it over to other PCs. Which is generally something you can work around… on Windows. Plugging this thing into my Steam Machine and having it default to forward and back messed up the bindings across games. There are probably ways to fix this—for example, there’s an app called Solaar that purports to manage Logitech devices on Linux. But for most users, this is a lot of hassle to deal with. And to be fair, if my goal is a single mouse for a single computer across multiple uses, moving to other devices shouldn’t factor that much in my evaluation anyway. Options+ is a passible way to create custom bindings for games, so long as you don’t need complex macros. I also wasn’t able to incorporate the MX Master 4’s haptic vibration into games. I suspect this is possible, but as haptic gaming mice have been tried before and abandoned as a concept, the juice probably isn’t worth the squeeze. Logitech is looking into it, for whatever it’s worth, just not on the MX series. But how does it actually play? Here we come to probably the most important aspect: How does it feel to play games with the Logitech MX Master 4? I tested it out in a series of games that needed fast, twitchy movement, some of which I’d played before and some of which I was trying for the first time, to evaluate how it functions as a mouse upon which a user might rely for all aspects in which they don’t switch over to a controller. The verdict: It’s fine. I was able to accommodate for the mouse’s larger size and weight without too much difficulty. The side buttons I was relying upon for special actions in games like Space Marine 2 and Far Cry 3 performed pretty much exactly as you’d expect. The 8000 DPI is far more than I usually use—2500, if you’re wondering—and I never experienced any game-breaking input lag with the USB wireless receiver. Saber Interactive I was expecting to have the most trouble when trying out new games. Nope! When playing Anger Foot, which is essentially a 3D, first-person version of Hotline Miami, I was able to move around the frantic small spaces and pinpoint shots without problems. I can see two possible exceptions here, and both of them are for gamers who are a long way away from my general style. If you really do need over 20,000 DPI, and if you need an insanely light mouse like, say, the G Pro X Superlight, you’re going to hate trying to use the MX Master or anything else like it. Michael Crider/Foundry I generally like a heavy mouse. The Naga V2 Pro I use is 131 grams according to my kitchen scale, for example. But even so, the MX Master 4 (150 grams) feels big and sluggish in my hand. Don’t get me wrong, it’s comfy, and it’s not as if it’s unmanageable. But it is two to three times as heavy as today’s high-end gaming mice, even the wireless ones. Fine for most, but competitive gamers look elsewhere So there ya go. I feel like most gamers would be fine with Logitech’s $120 MX Master 4, with the exception of those who get competitive for shooters or games like Marvel Rivals. You’re going to feel that weight. And for other games that need complex macros, or who want to keep the same settings across multiple devices, the software limitations might get you down. If you fit into either of those camps, you’ll have an easier time going with a dedicated gaming mouse and adapting it to non-gaming uses. But otherwise, the Logitech MX Master 4 pulls off some surprisingly impressive gaming chops. Cool stuff from around the web CyberPowerPC is making LEGO-compatible PC cases: So far they’ve only shown up in one anime expo, at least as far as I can tell. So who knows if these neat designs will actually be sold. Microsoft blames latest Windows 11 crash on RGB lights: Yet another strike against turning your PC into a disco ball, the company says that games including Arc Raiders, Marvel Tokon, and The Finals are crashing because of RGB drivers. On the other hand, Noctua says it’s not interested in RGB. It looks like the iconic brown for the cooling component supplier isn’t going anywhere. Power supplies with removable, modular ports: I freakin’ love this idea. Thermaltake’s new PSU designs let you remove the entire connection plate, plug everything in, and then you have just one thing to connect inside the case. Genius. Thermaltake Hey, maybe don’t download a pirated copy of GTA VI. I’m sure a lot of you are looking forward to the biggest game evah (even if it won’t be on PC for a year or two), and know there’s a build floating around out there. Don’t try to download it, of course it’s full of malware. Game pick of the week: Terror of Hemasaurus Loren Lemcke It’s been a while since we had a good Rampage game, hasn’t it? Actually it’s only been about four years, because this game is the spiritual successor to the classic monster bash-em-up. And also South Park, because it has a story too, which is even more cynical, jaded, and self-aware than Rampage was when it was gracing consoles in the 90s and 2000s. Terror of Hemasaurus is all about smashing cities, fighting the military, and crunching down on tasty humans. The Rampage formula still works great, and with more speed and creative movement options it’s even better. If you’ve never kicked a mouse-sized human into a helicopter, you simply must try it. The game is short and sweet, but you can play it with up to four friends in local co-op. And you should. Just make sure everyone’s over the age of 13 or so. Terror of Hemasaurus is $15 on Steam. It’s also available on all consoles. That’s all for this week. To get my editorials, news picks, and game recommendations delivered straight to your inbox, you can sign up for Boost Mode on the PCWorld Newsletter hub. Also check out The Full Nerd, Prompt Mode for “AI” tips, Safe Mode to keep your PC and gadgets secure, and Smart Mode to boost your productivity.

为什么现在还能做

Based on this period’s screening signals

建议怎么做

Recommend verifying with the original source and running small-scale tests

变现说明

Monetizable via productized services, subscriptions, or consulting

可复制性

Further replication feasible once demand is clearly established

机会视角：Opportunity to explore replicable products or services around this signal

信号

Johannesburg: As SA20 builds towards its milestone fifth season, the independently assessed economic impact of Season 4 has underlined the League's growing contribution to South Africa, generating $406 million in GDP contribution and creating or sustaining almost 9,500 jobs across the country. The final Season 4 Economic Impact Assessment (EIA) records the League's strongest economic performance to date, with increases across direct expenditure, GDP contribution, employment and household income. Four seasons in, and the League's biggest numbers aren't on the scoreboard. #BetwaySA20 Season 4 delivered a R6.5 billion boost to the South African economy, our strongest season yet. A 23% jump year on year, almost 9 500 jobs created or sustained across the country, and R1.7… pic.twitter.com/BpGDOso5hS — Betway SA20 (@SA20_League) August 25, 2026 A record-breaking Season 4 saw attendances increase by 13 per cent, with 22 of the 34 matches selling out across the six host cities. The Final sold out in record time for the fourth consecutive season, further establishing SA20 as one of the marquee events of the South African summer. The growth in attendance and activity surrounding the League helped drive direct expenditure to $137 million, up from $112 million in Season 3, which translated into a $406 million contribution to GDP, representing a 23 per cent increase year-on-year. The League's contribution to employment also grew significantly, with 9,470 annualised jobs created or sustained across South Africa, up 16 per cent from Season 3, while household income increased from $87 million to $106 million. Player registration for the #BetwaySA20 Season 5 Auction is open! Registrations open today, 24 August 2026, and close at 17h00 (SAST) on Monday, 14 September 2026. Register now https://t.co/NkmsGXilyr pic.twitter.com/7sjuOLsn5Z — Betway SA20 (@SA20_League) August 24, 2026 League Commissioner Graeme Smith said the independently assessed results demonstrate how the League's growth is increasingly translating into wider economic value. "Four seasons in, we can see the impact of Betway SA20 extending well beyond what happens on the field. A $406 million contribution to GDP, almost 9,500 jobs supported and $106 million in household income are significant numbers and demonstrate the broader value that a successful South African sporting property can create,” Smith said. "As we head into our fifth season, our ambition is not only to continue growing the League, but to deepen that impact across cricket, our host cities and the wider South African economy." The assessment also reflects the economic activity generated across SA20's six host cities and the role of its six Indian Premier League (IPL)-backed franchises in driving investment, attendance and activity in their respective markets. 'Don't Be Scared, You're Captain': Sarfaraz Khan Taunts Dhananjaya De Silva In Viral Sledging Battle During IND Vs SL 2nd Test | VIDEO "What is particularly encouraging is that this impact is being generated across the country," Smith added. "Our six franchises and host cities have been fundamental to the League's growth from day one. Every full stadium brings people into these cities, supports local businesses and creates opportunities around the event. As the League grows, we want that contribution to grow with it." Season 4 also saw SA20 continue to expand its social impact footprint. The League raised $31.2k for social impact partner Laureus Sport for Good Foundation and provided more than 400 young Laureus beneficiaries with the opportunity to experience live cricket. The Season 4 results provide a strong foundation as SA20 prepares for its fifth edition, beginning on 17 January 2027. The build-up to the milestone season will gather momentum with the Season 5 Player Auction on 7 October, ahead of another summer featuring some of the world's leading T20 players. Fans can register now for the Season 5 early-access ticket pre-sale at sa20.mobi/s5-ticket-presale-registration, giving them first pick of seats before tickets go on general sale, including tickets for the Playoffs and Final. (Except for the headline, this article has not been edited by FPJ's editorial team and is auto-generated from an agency feed.)

为什么现在还能做

Based on this period’s screening signals

建议怎么做

Recommend verifying with the original source and running small-scale tests

变现说明

Monetizable via productized services, subscriptions, or consulting

可复制性

Further replication feasible once demand is clearly established

机会视角：Opportunity to explore replicable products or services around this signal

信号

Macworld - Review roundup of the top Thunderbolt and USB-C docks for MacBook Air, Pro, and Neo in 2026.

为什么现在还能做

With the adoption of Thunderbolt 5 and new MacBook lineups (including M5 chips), 2026 marks a pivotal year for upgrading docking station hardware to match new bandwidth and display capabilities.

建议怎么做

Users should identify their MacBook's specific ports (TB5 vs TB4 vs USB-C) and match them with docks that support their required number of external displays and peripheral needs.

变现说明

Affiliate commissions from product sales and ad revenue from high-intent tech traffic.

可复制性

Yes, this is a classic 'best X for Y' comparison format that can be replicated for other tech categories or ecosystems.

机会视角：Positioned as a definitive buying guide for professionals and Mac users needing to expand connectivity and productivity setups.

信号

Medical awareness and telemedicine marketing targeting parents and adults suffering from dyshidrotic eczema (sweat herpes) symptoms during hot weather.

为什么现在还能做

Seasonal spike in interest for skin issues related to heat and sweating; highlights the convergence of health anxiety and digital healthcare solutions.

建议怎么做

Content addresses a specific pain point (itchy blisters) and funnels users to online dermatology consultations for diagnosis and treatment.

变现说明

Lead generation fees or revenue share from telemedicine platforms (Dxy/Dingxiang Mom).

可复制性

Yes, applicable to seasonal health trends and symptom-based queries in other regions.

机会视角：Health education coupled with direct response marketing for telemedicine services.

信号

Launch and promotion of DW Access, a sideloaded news app designed to bypass internet censorship and surveillance in restricted regions.

为什么现在还能做

Growing global concerns about internet freedom, government surveillance, and the tightening of app store controls in certain geopolitical climates.

建议怎么做

Provides technical guides on how to sideload the app, explaining its security features (VPN, Hash ID) and addressing user fears about monitoring.

变现说明

Non-monetary; supports the media organization's mission of providing uncensored news.

可复制性

Yes, similar models exist for other news outlets or privacy-focused apps needing to bypass platform restrictions.

机会视角：Information freedom advocacy and technical support for circumvention tools.

信号

Nostalgia trend triggered by 'Post-00s' (Gen Z) now being working adults; social media campaigns re-engaging childhood cartoon characters.

为什么现在还能做

Generational milestone: the first wave of Post-00s are entering adulthood, prompting a cultural reflection on childhood innocence versus adult pressures.

建议怎么做

Emotional storytelling content that validates the audience's feelings of stress and offers nostalgia as a source of comfort and resilience.

变现说明

Brand engagement, ad revenue, potential for merchandise or licensed content partnerships.

可复制性

Yes, nostalgia marketing is highly replicable across different generations (e.g., Millennials remembering 90s toys).

机会视角：Emotional resonance marketing and cultural commentary.

信号

Pre-release rumors and leaks regarding Apple's 20th Anniversary iPhone, expected in 2027.

为什么现在还能做

Apple has not released a dedicated 'Anniversary Edition' since the iPhone X in 2017; anticipation builds as the 2027 date approaches.

建议怎么做

Aggregates rumors on specs (display, battery, camera), design changes, and release timeline for a highly anticipated product cycle.

变现说明

Affiliate links for pre-orders; ad revenue from high-volume tech traffic.

可复制性

Yes, standard practice for tech media to cover major product launches and anniversaries.

机会视角：Tech rumor analysis and consumer guidance for potential upgrades.

信号

Have you or your child ever had these small blisters on your hands or feet, especially prone to recurrence in summer?

为什么现在还能做

Compiled based on this period's screening signal

建议怎么做

Suggested to verify with the original text and test on a small scale

变现说明

Can monetize through productized services, subscriptions, or consultations

可复制性

Can be further replicated once clear demand is identified

机会视角：Explore replicable products or services around this signal

信号

DW News. Why develop this new app? Currently, the world faces multiple crises, one being increasing government and big tech control over the internet.

为什么现在还能做

Compiled based on this period's screening signal

建议怎么做

Suggested to verify with the original text and test on a small scale

变现说明

Can monetize through productized services, subscriptions, or consultations

可复制性

Can be further replicated once clear demand is identified

机会视角：Explore replicable products or services around this signal

信号

A warm wave of nostalgia recently swept social media platforms: domestic animation accounts used classic characters to converse with adults born after 2000.

为什么现在还能做

Compiled based on this period's screening signal

建议怎么做

Suggested to verify with the original text and test on a small scale

变现说明

Can monetize through productized services, subscriptions, or consultations

可复制性

Can be further replicated once clear demand is identified

机会视角：Explore replicable products or services around this signal

信号

Macworld. The original iPhone was released on June 29, 2007. So next year, it will turn 20 years old, and rumors point to Apple commemorating the event with a special 20th-anniversary model.

为什么现在还能做

Compiled based on this period's screening signal

建议怎么做

Suggested to verify with the original text and test on a small scale

变现说明

Can monetize through productized services, subscriptions, or consultations

可复制性

Can be further replicated once clear demand is identified

机会视角：Explore replicable products or services around this signal

信号

Tasty! Super tasty! It's the kind of wine that requires no knowledge of varietals, winemaking processes, tannins, structure, balance, or vintage.

为什么现在还能做

Compiled based on this period's screening signal

建议怎么做

Suggested to verify with the original text and test on a small scale

变现说明

Can monetize through productized services, subscriptions, or consultations

可复制性

Can be further replicated once clear demand is identified

机会视角：Explore replicable products or services around this signal

信号

2026年是雍禾2.0模式落地的元年——从过去粗放式的流量驱动，转向以精细化运营、医疗能力和客户长期价值为核心的增长模式。养固业务的崛起，正是这一模式转换的集中体现。 作者：王熠礼封图：图片资料室 8月20日，雍禾医疗发布2026年上半年财报：营收9.81亿元，同比增长13.8%；净利润7550万元，同比增长170.9%，半年净利润已超过2025年全年。 拆解这份财报，增长的结构比增速本身更值得关注。植发业务收入7.43亿元，同比增长12.3%，保持稳健；医疗养固业务收入2.27亿元，同比增长20.7%，增速显著快于植发。 2026年是雍禾2.0模式落地的元年——从过去粗放式的流量驱动，转向以精细化运营、医疗能力和客户长期价值为核心的增长模式。养固业务的崛起，正是这一模式转换的集中体现。它不再是植发的附属品，而是雍禾从“植发机构”向“毛发健康管理平台”转型的关键载体，也是检验其商业模式能否从低频手术延伸至高频服务的重要标尺。 雍禾的成绩获得了资本市场的认可。中金公司在研报中指出，雍禾医疗已实现从“规模增长”向“质量增长”的策略转型，正构建起长期技术护城河，机构维持“跑赢行业”评级并将目标价上调。 从“一锤子买卖”到“全周期管理”，雍禾如何打开行业天花板 植发解决了“已经秃了”的问题，但2.5亿脱发人群中的绝大多数，需要的不是手术，而是早期干预和长期管理。国家卫健委数据显示，中国脱发人群已达2.5亿，国内植发渗透率仅约0.2%——这意味着，植发只是一座冰山露出水面的那一小部分，真正庞大的市场在水面之下。 传统生活级养发馆以按摩、护理为主，缺乏医生专业诊断，服务周期长、效果难以量化，对脱发的干预停留在“养护”而非“治疗”层面。而有专业诊断能力的植发机构，过去只做手术，不做长期管理。两者之间，存在一个巨大的市场空白。 雍禾找到的突破口，正是这个空白地带——用医疗级能力介入养固，把低频的植发手术变成高频的长期服务。这不仅是业务边界的拓展，更是对整个行业“天花板”的结构性突破。当一家公司从“做完手术就结束”转向“覆盖脱发全生命周期”，它打开的不仅是收入结构，更是行业的想象空间。 在消费医疗行业普遍承压的上半年，雍禾的增长显得尤为突出。国家“十五五”规划明确提出推进健康中国建设，将医疗健康服务列为重点发展方向。毛发健康作为消费医疗的细分赛道，正在从“小众需求”走向“大众消费”。弗若斯特沙利文联合头豹研究院发布的《2024年中国毛发健康服务行业研究》报告显示，中国毛发健康市场预计到2028年将达到1165.4亿元。行业扩容叠加政策利好，为率先完成从“手术交付”向“全周期管理”转型的企业，提供了足够宽阔的增长赛道。 从“配套服务”到“独立业态”，医疗级养固的标准化之路 医疗养固并非新概念。雍禾早在几年前就已布局“店中店”模式，在每个植发机构中设立史云逊医学健发中心。但当时的养固更像是植发的“配套”——术后维护、客单价补充。 2026年的变化在于：养固正在从配套走向独立。 首先是定位的升级。雍禾将养固业务定义为“医疗级”，与市面上面向生活美容场景的传统养固业务形成本质区隔，核心差异在于三件事：医生诊断作为服务的起点，医疗级器械与药物作为干预手段，以及可量化、可追踪的效果评估体系。 在增长逻辑上，养固业务的收入提升来自两个层面：一是产品矩阵的丰富与升级，今年胶原多肽、微针等配方的推出提升了套餐认知度，当期复购金额同比增长59.7%，复购客单价集中在12000元以上；二是消费者教育的深化，通过内容科普和植养一体优惠活动，消费者逐步认识到植发与养护的协同关系，植养联合患者人数较去年同期明显提升。养固不再是“术后顺便做一下”的配角，而是消费者主动选择的一站式解决方案中的重要环节。 雍禾正在推行一种“54321”医养闭环模式：5种疗法（配方渗透、多肽、胶原、光电、微针美塑）、4个养护阶段（头皮调理→毛囊修复→营养增粗→维养抗衰）、3重产品体系（配方产品、中胚产品、药品，院用+居家覆盖）、2大场景（到院治疗与居家养护）、1对1专属管理师。这套体系将此前高度依赖个体医生经验的养固服务，转化为可复制、可标准化的产品。 其次是数字化能力的注入。雍禾自研的图像采集设备已应用于术前评估与术后追踪，3D智能采集设备完成预研。2026年上半年，AI智能检测报告单月生成量超5万份，AI图像识别技术覆盖全国院部。AI医疗模型实现门诊病历智能化，准确率超95%。这些技术将此前依赖于医生主观判断的诊疗过程，转化为可量化、可追溯、可分析的数据流，在支撑医疗服务标准化的同时，也为养固业务的规模化复制创造了条件。 在渠道端，2026年9月，深圳首家“雍禾毛发健康”小型诊所将开业。这是雍禾“1+N”战略的落地起点：一个手术中心布局N家小型毛发诊所，小店不做植发手术，只承担诊断、医疗养固和非手术类治疗。单店面积约200平，一次性投入约120万元，月度盈亏平衡点约30万元。 “1+N”用轻量化的小店模式降低了扩张门槛，将养固从植发的附属服务变成了独立的流量入口。小店可以更灵活地渗透社区和商圈，覆盖轻度脱发人群和术后长期管理人群，实现“高频触达、低客单价、高复购”的运营闭环。 按照规划，雍禾3年内将实现80家手术中心、150家皮肤诊所的网络布局。植发机构不再单点作战，而是以“中心+卫星”的方式形成覆盖网络。养固的独立化，本质上是渠道模式和商业逻辑的双重重构。 养固如何成为雍禾的“慢变量”护城河 如果说植发是解决“已经秃了”的问题，养固解决的是两件事：一是“还没秃但快秃了”的早期干预，二是“植完了以后怎么维护”的长期管理。前者是入口，后者是存量运营。两者的结合，构成了“获客—服务—复购—长期管理”的商业闭环。 2026年上半年的数据已经显现闭环的初步效果：医疗养固服务人数56,757人，同比增长26.4%，复购率25.9%，复购金额同比增长59.7%，复购客单价集中在12000元以上。这些数字意味着客户不是“来一次就走”，而是持续产生价值。 而外部环境正在为这个闭环提供新的变量。全球脱发药物市场正处在爆发前夜。JAK抑制剂、雄激素受体拮抗剂等创新药物陆续获批或进入临床后期。市场机构预测，全球脱发药物市场规模将从2025年的90.9亿美元增长至2034年的126.3亿美元。药物是“医疗级养固”的核心工具之一，但药物需要与专业诊断、方案设计、长期跟踪管理结合才能最大化价值——这正是医疗机构而非美容机构的优势。 雍禾的另一个布局也值得关注。史云逊正从健发门店品牌转型为生物科技公司，布局药品、特妆、妆字号产品的研发生产，未来面向B端和C端提供产品。这意味着养固业务的边界从“服务”延伸到了“产品”。当脱发药物市场真正放量时，史云逊的产品管线有可能成为雍禾在养固领域的差异化武器——不仅提供服务，也提供“服务+产品”的组合方案。 从中长期看，养固业务的门槛不在于设备或产品，而在于三件事：是否有足够规模的真实诊疗数据支撑方案优化，是否有标准化的服务体系和医生培训机制保障质量，是否有覆盖广泛的服务网络承接需求。 过去的植发行业是“流量生意”——广告砸出去，手术做掉，关系结束。现在雍禾正在尝试的，是以医疗能力为底座，以养固业务为纽带，把客户变成用户，把单次交易变成持续服务。 从植发机构到毛发健康管理平台，从低频手术到高频服务，从流量驱动到效率驱动，雍禾的这条路径正在被数据逐步验证：毛利率从63.9%提升至69.0%，EBITDA同比增长29.2%，经营现金流净额达2.64亿元。效率提升带来的盈利改善，不是一次性的，而是结构性的。雍禾在多个维度上积累的“慢变量”构成了竞争对手短期内难以复制的护城河。这不是靠砸钱买流量能快速追赶的，而是靠时间沉淀出来的体系能力。 2.5亿脱发人群、0.2%的植发渗透率、正在爆发的脱发药物市场——这三个数字指向同一个结论：毛发健康行业的竞争，才刚刚进入“下半场”。上半场拼的是流量捕获，谁砸钱多谁跑得快；下半场拼的是存量运营和医疗深耕，谁留得住用户谁才能活得久。 雍禾从“植发机构”向“毛发健康管理平台”的转身，正是应对这一变局的关键落子。当植发从“一锤子买卖”变成“全周期服务”的入口，当养固从配角走向独立业态，雍禾构建的已不是单一的手术交付能力，而是一套覆盖诊断、治疗、养护、复购的完整闭环。 阅读原文 平台地址： http://www.jintiankansha.me/t/7NVsbeb2af

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

今天的文章，我删删改改了很多遍，一直在犹豫要不要写出来。 但这事太重要了，每个人都可能会遇到，还是决定写出来想和大家分享，希望大家看完后能都重视起来，引起警觉。 几天前，好友打来语音，说家里出了事，想找我借点钱。 他说，不是不得已，自己是张不开这口的。主要是家里的老母亲突发心肌梗塞，砸了十几万，人才好不容易从ICU中抢救回来。 医生说，想保证今后的生命质量，还得做心脏搭桥才行，至少要再付出几十万。 在我的印象里，他的条件还可以，前几年买了房，按理说不至于到找别人开口借钱的地步，这次主动开口，说明是真的遇到难处了。 后来才知道，他在的公司年初解散了整个团队，没有进账不说，还得雷打不动还一大笔房贷、养一大家子人。 朋友巨大的压力，我非常理解。作为多年老友，我很想去帮助他，但心里又很纠结。借钱吧，知道治病是个无底洞，还上这笔钱遥遥无期，自己也在勒紧裤腰带，日子过得紧巴巴。不借吧，心里又实在是过意不去，今天四处求钱的是他，明天可能就是我。 最后拧巴了半天，还是转了些块钱过去，想让自己内心好受点。 说实话，到现在我都没缓过来。 这些年也不是头一次被朋友借钱，只是突然发觉，人到中年，一头是亲生骨肉，一头是生养自己的父母。 很多人生活看似光鲜亮丽，其实每一步都如履薄冰，一点风吹草动就可能掉进冰窟里出不来了。 即便是老师、公务员这种大家都认可的稳定、体面的工作，也随时会遭遇巨大的风险。 江苏的吴先生，39岁，当了半辈子老师，安守本分，兢兢业业。而就在前几天突然出现在论坛上，请网友投票决定父亲的生死。 他父亲骑电动车不慎摔倒，没想到一次小意外竟造成了严重后果，脊椎受损、肺部感染。 从家乡转去省会城市的医院，做完两次手术休养半个多月后，父亲终于慢慢好转。 可是一家人刚出院回到家中，父亲的伤口却又开始化脓，同时出现颅内感染、脑脊液漏、切口感染等问题。 父亲不得不再次住进ICU，一天花费少则两三千，多则五六千。 医生告诉吴先生，如果保守治疗，短期内他父亲就可能有性命之忧，有可能下周末都撑不到。 如果做开颅手术，康复的几率也很小， 一百万花下去也不排除人财两空的结果。 虽然吴老师和妻子都有工作，父亲也有医保。 但医保能保的部分十分有限，加上后续未知的庞大治疗费与康复费用，这无疑是压在这个普通双薪家庭头上的一座大山。 为了省钱，吴老师六七十岁的老母亲甚至不愿意去住旅馆，一直在医院长椅上过夜，看着老人一把年纪还要受苦，作为儿子他实在是心疼。 但他除了是儿子，还是丈夫和父亲。 如果倾家荡产，举全家之力也没保住父亲， 那么活着的孩子、妻子和他就要长期负债，承受居无定所，艰难度日的绝望。 种种考量，万般纠结，迟迟下不了决心的吴先生决定发起投票，将父亲的生死交给网友。 毫不意外，网络一片骂声：还没尽全力就让一群陌生人来决定亲爸的生死？这配当老师？ 吴先生配不配当老师我不做评价，但凭借我多年的社会阅历， 在“老人和小家之间怎么选”这个世界难题上，大多数人其实都会选后者。 孩子患病，家长大多会砸锅卖铁、不惜一切，而年迈父母患病，成年子女首先权衡的是钱，是现有生活是否会受影响。 因为当上有老、下有小时，“救不救”不只是良心问题， 还是一个有限资源的可持续分配问题。 试想，若能负担得起，谁会在老人与小家之间徘徊犹豫？谁会纠结于计算钱与生命哪个价值更高？ 没进过医院的你，根本不知道钱有多不经花。 拿最常见的癌症来说，住院费、医药费、护理费、设备使用费等，随便一圈下来可高达几十甚至上百万， 轻易就能将一个家庭几代积蓄全部没收。 如今的我们，万事顺遂的时候，一切看起来都还过得去。但意外一旦来临，大多都招架不住。 不敢想，万一哪天我倒下了，年迈的父母如何安享晚年，孤身的伴侣如何支撑整个家庭？ 我们很难避免一辈子不遇到这些事情，也无法避免它给我们的生活带来冲击。但是有两样东西是我们可以决定的： 第一，是面对苦难的态度。 第二，是未雨绸缪的准备。 至于如何准备？思来想去，发现其实也没太多的办法，就两条路： 1、在利用好杠杆的同时，还要不断增加收入、控制开支，掌握尽可能多的存款，以应对可能出现的意外。 只是放在今年这个行情里，收入能稳住就已经不易，指望大幅增收并不现实。这条路对于大多数人来说，都是越来越窄。 2、对于结余比较少、收入增幅比较慢的读者，可能最好的办法就是把自己和家人的保险配置齐全，用尽可能少的钱，把自己不能承受的风险转嫁给保险公司。 （本文不推荐任何保险产品） 关注我的老粉丝都知道，一直以来我都在建议大家一定要把保险配置齐全，那是因为，这些年我在医院里见证了太多这样的悲剧， 而这些悲剧很多时候都是一份很便宜的保险就能解决的事情。 借此机会，想再给大家做点简单的保险科普，都是很实用的知识，希望大家耐心看完。 相信大部分读者都已经参加了城镇医保或者新农合，因而可能会觉得已经后方无忧，那你就错了。 虽然国家这些年一直在医保方面做了不少努力，每年的灵魂砍价都能登上热搜。但在感谢国家的同时我们也要认识到，国家的医保只是基础医疗保障，毕竟并不是万能的。 比如，医保可报销的药品有3200多种，在市面上15多万种药物中只占很小一部分；而且在医保起付线以内、封顶线以上、报销比例外和医保目录外的医疗费用，都是要自己承担。 保险的作用就是在遇到疾病、意外等风险时，帮我们承担风险，把损失降到超低。 假设一场病可能要花费80万，经过社保报销和医疗重疾险赔付后，没花一分钱，剩余100万拿来补偿患病后5年家庭收入损失， 生场病也没影响家人正常生活。 关注我的读者应该很多都是家里的经济支柱，我相信，动辄几十、上百万的治疗费并非所有人都能轻易掏的起的，就算掏得起，那又得花多少年来追回这笔损失？ 可以这么说，没保险的家庭就像一个脆弱的玻璃球，遇到冲击整个家庭就碎了。而有保险的家庭就像是一个皮球，即使被打压也能够弹起来。 但一定要注意，保险不能瞎买。 保险本身是有很高的门槛的，涉及金融、医学、法律三方面知识交叉。 因为不懂相关知识，还有部分业务员为了提成进行销售误导，以致于很多人买贵、买错，甚至最后赔不了钱的，大有人在，大家见过的也不止一个两个了。 如果有一家专业平台，能清清楚楚告诉我们究竟买了啥、能保啥、该花多少钱合适，避免大家掉进买错保险的坑，那对我们的帮助会非常大。 几番对比之下，还是觉得老朋友大童最靠谱： 大童成立于2008年，是一家全国性保险服务机构，老牌的正规军。 2020年还跻身世界保险中介前20名，是全亚洲唯一入围的保险中介企业。 它不属于任何一家保险公司，是专门做服务的。不会未了解具体情况，上来就推销产品，而是站在我们的立场，从140+家保险公司的产品里，帮忙筛选适合的产品。 大家在挑选产品前，都可以去找他们帮忙对比和挑选。 我体验过他们的保障规划服务，顾问会先了解我的基本情况，询问我的需求，例如： 之前有购买过哪些保险产品？ 家里几口人？家中的经济支柱是谁？ 想通过保险解决怎样的问题？ 家庭可支配收入以及保险预算有多少？ 然后根据我的 家庭结构、财务状况、保障需求， 站在我的角度，帮忙规划一份详细的家庭保障方案。 是真正的1对1量身定制，相比传统机构还能节约 30%-50%的费用。 （保险顾问详细沟通后给出专属的方案） 拿到方案后，顾问还一对一帮我讲解，确保我能深度了解方案，对自己家庭的保障规划更清晰。 真心希望大家都能去预约下这项服务，没有保险的朋友，能知道 应该买什么、买多少额度 才能解决家庭的问题，以及 花多少钱才合理 ； 已经买了保险，顾问也可以帮忙分析已有保单，帮忙看看买的值不值，方便及时 查漏补缺 。 我特意争取了 50个免费咨询的名额 给到大家，先到先得，可长按扫描二维码报名领取。 老毛粉丝专属福利 长按扫码报名 即可 免费 获得价值 500元 专属家庭保障规划方案 👇👇👇 ▲ 长按识别二维码报名 前50名免费 【可升级】高端资产配置服务 大童还有独特的 线上服务+线下咨询 的能力，全国线下服务网点200+个，需要时有专人跟进理赔，省时又省心。为大家特别争取了 50个免费咨询名额 ，建议大家都趁这个机会，为孩子和家人做一份完整的保障规划。另外，这次报名成功的小伙伴，还给大家争取到了 1份额外专享福利！ 免费获得全网首发 人手必备的 《家庭保障规划宝典》 电子版一份！ 这份地图只送不卖，包含14+个模块、41+张图表和300+个知识点，简单直观，有趣易懂。 早规划早安心，不仅是远见，更是对自己和家人的负责！ （长按识别下方二维码领取专属福利） 老毛第五次倾力推荐 ， 限前50免费名额 手慢无！报名如下 👇👇👇 （长按识别二维码领取， 前50名免费 ）【可升级】高端资产配置服务 文章原文 平台地址： http://www.jintiankansha.me/t/UrRhyXSGRF

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

近年来，高尔夫正越来越多地出现在中国高净值家庭对小孩的教育规划里，这项曾经贴着“小众”标签的运动，正在成为他们的家庭运动项目。 据中国高尔夫球协会《中国青少年高尔夫球发展报告（2025）》，截至 2025 年底，全国累计注册青少年球员已超过 15.4 万人——而 2014 年，这个数字还只有 400 人。 人多了，比赛也多了，更多孩子站上了发球台。高尔夫究竟凭什么，能在一众运动里脱颖而出，走进这些家庭的育儿清单？ 为什么说高尔夫是一项理想的亲子运动？ 在竞技体育里，高尔夫或许是亲子属性最强、最独特的运动——它不只是简单地一起玩，它的运动特点、整个赛事体系的设计等等，天然就适配亲子的场景和目的。 依托高尔夫特有的差点计分规则，家长和小朋友能在同一个体系下认真较量，当小朋友在这样的环境中战胜成年人，那份成就感是其他亲子运动给不了的。而一场球通常持续 4 到 5 个小时，家长和孩子边走边聊、讨论战术，远离手机和外界干扰，这是真正属于彼此的深度沟通。在亲子沟通之余，高尔夫还自带品格教育的底色。自我报杆、修复草坪、独立决策并为选择负责……球场的规则天然承载礼仪，家长示范，孩子模仿并内化，品格教育在潜移默化中发生。 12 岁李咏轩的高球之路源自爸爸，最近一场父子球局是上个月在风神球场，虽然现在爸爸已经挥不动咏轩用的那支球杆，但老将依旧凭借稳定的心态与儿子打得胜负难分。力量强弱的悄然转变并不改变两人的交流，父子俩聊战术、聊心态调整。高尔夫真正的妙处，是它把亲子相处拉长成了一段持续进阶的陪伴。 爸爸经常告诉咏轩：“要有赢球的本事，更要有不怕输的胆量。成绩不理想时，要敢于面对，及时复盘，总结经验，再努力去赢。球场就是小小的人生，跌倒不可怕，怕的是没有重新站起来的勇气。“ 同样 12 岁的徐逸菡，她的高尔夫故事是从“试试看”开始的。高尔夫刚进入这个家庭时的定位很单纯，就是一项亲子运动。但随着逸菡逐渐从高尔夫中学会了专注，开始斩获成绩，这项运动的角色也从家庭活动转变为学业规划路径中的一环。高尔夫的角色变了，但家人的陪伴没有改变，每场比赛，妈妈都在她身边，是她的专属“球童”，陪伴鼓励她一路向前。 如果问高尔夫带给了孩子什么？徐逸菡妈妈会说“有很多”，在她看来，孩子付出的努力和汗水，回报的绝不仅仅是成绩单上的数字、排名板上的名次，而是“孩子由内而生的自信”。 从最开始在练习场，长辈、球童、教练，对她每一点小进步毫不吝啬的鼓励与肯定；到赛场上打出人生第一个红字，拿下属于自己的第一座冠军奖杯；再到听见她自我调侃式地肯定自己，“我真是个天才”。这份自信，不是盲目自大，是一次次挥杆、一场场比赛，靠自己拼搏换来的底气。高尔夫给她的，不只是球技，更是相信自己的力量。 两个家庭，两种不同的收获。对于孩子们来说，高尔夫教会他们专注、冷静，而专业的赛事体系也给他们提供了课堂上难以习得的内在力量。对父母来说，高尔夫也悄然改变了亲子陪伴的方式与内涵。随着高尔夫正逐渐成为高净值家庭教育的新趋势，在这股趋势背后，早已有一家品牌在这一赛道深耕布局。 迪桑特儿童把赛事级的专业和仪式感融入了成长的日常 如果把视野放到赛场，会发现迪桑特儿童在高尔夫领域的投入，远比想象中的更深，也更持续。 作为国内最具影响力的青少年赛事体系之一，自 2007 年“中高协-汇丰中国青少年高尔夫发展计划”启动以来已走过 20 年，让超过 14 万名儿童接触到了高尔夫。而从 2024 赛季开始，迪桑特儿童作为汇丰青少年发展计划官方赞助商及官方服装赞助商，为小球员们提供专业装备并围绕赛事构建了一整套以“星”为主线的荣誉体系，将仪式感融入赛事日常，陪伴高球少年成长。 上周，2026 年汇丰全国青少年高尔夫球冠军赛安吉站在浙江安吉龙王溪乡村俱乐部启幕。开赛当日，迪桑特儿童年度最具价值球员李咏轩、徐逸菡来到现场，作为冠军球员代表带着父母的祝愿，踏上了全“星”征程——而他们的故事，正是迪桑特儿童构建的“星”体系最好的注脚。 以“星战袍”为起点，品牌为每一站获得赛事冠军的小球员在赛事服上烫印星星，用星星印记见证每一次夺冠。同时围绕年度赛事，让“开赛仪式”成为记忆锚点，“耀星”仪式记录青春荣耀，“星荣耀”则为年度最具价值球员加冕，品牌更会为年度最具价值球员提供与顶级球员面对面交流的机会，将优秀小球员引向更高更大的平台。一年三个重要的仪式记忆，让父母一同见证孩子的荣耀时刻，逐渐成为汇丰赛事中家庭共同记忆的一部分。 而在“星战袍”的背后是一条完整的儿童高尔夫专业装备线。从高尔夫球场真实环境出发，面料与版型剪裁皆致力于适应球场环境且助力挥杆无束缚，同时品牌也充分利用赛事机会，与球员复盘穿着的真实体验，持续在产品上带来迭代升级。对迪桑特儿童来说，赛事不只是竞技的舞台，也是产品被反复检验的现场，而这套以赛代测的研发体系，本身就是专业性最可靠的来源。 9 月即将上新的 EV JACKET 也从秋季球场环境多变考量出发，外层面料防风防水，能抵挡凉风和突发小雨；内里的弹力抓绒轻便保暖，全贴合工艺让防护与活动互不妨碍，挥杆依然自在。 品牌的专业陪伴也从赛场延续到日常。围绕汇丰青少年赛事，在品牌面向用户的私域社群“小行动家俱乐部”里也打磨出了“冠军课堂”、“冠军挑战赛”、“冠军爬藤之路”三个成熟项目，搭建起一套从赛场延伸至校园与社群的完整活动体系。 其中的核心项目“冠军课堂”把汇丰青少年冠军请进校园，与同学们面对面交流成长故事，从 2024 年至今这一项目已经在 10 座城市的 10 所国际化学校开展，李咏轩和徐逸菡都曾作为冠军球员参与其中，与各地小行动家分享自己的高球故事。从练习场上的日复一日，到第一次站上赛场的心情——成长的轨迹被摊开细述，热爱这件事，在小小行动家面前变得具体可感。 秉承着 GOLF FOR FUTURE 的理念，迪桑特儿童希望高尔夫能成为陪伴孩子长期成长的一部分，通过专业装备支撑每一次上场，赛事体系给出一级一级向上的台阶，多元的运动体验让这项运动自然长进家庭的日常。 跻身奢品童装阵营，迪桑特儿童持续落地高端育儿生活方式 对品牌而言，专业与陪伴不止发生在赛场和社群，也延伸到了门店——这里既是选购专业装备的场所，更是感受高尔夫文化、参与社群活动的线下亲子体验场。 南京德基广场 2025 年全年销售额达 262.4 亿元，位居全国单体商业项目之首；在这样的空间里，铺位不只是物理坐标，更接近一种行业排序——谁与谁相邻，传递着商场对品牌的判断。近期，迪桑特儿童门店搬迁重装立于奢侈品童装比邻的铺面，儿童运动品牌走进这片阵容，在德基是第一次。 把运动体验装进育儿消费场景，门店便不只是消费场所，更成为感受运动文化、交流育儿理念的亲子体验场——它提供的不是一件衣服，而是一种被验证过的高净值家庭生活方式。从门店，到冠军课堂，再到果岭上的亲子时光，这些场景共同拼贴出高尔夫生活方式的一块块拼图，让这项运动不再只出现在赛场和课堂里，而是融入了家庭生活的日常。 运动正在成为家庭的生活方式标签，即便孩子不深度参赛，家长依然愿意为这套价值买单。迪桑特儿童将“装备 + 赛事 + 社群”做成组合包，家长买的不是一件衣服，而是一条看得见的成长路径。从“装备消费”走向“服务型消费”，这正是品牌定位的深层逻辑。 写在最后 总体来看，当下的家庭教育早已不局限于书本里的知识和单一指标。体能、品格、视野与陪伴，成为了越来越多家庭看重的、综合性的成长维度。 在童装生意里，品牌和家庭的关系通常是以收银台为终点，但在迪桑特儿童看来，一件衣服只是故事的开始——星战袍上的一颗星星，排行榜上的一个名字，冠军课堂里一次与冠军球员面对面的机会，果岭上一段全家人的共同回忆。装备、赛事与社群连成一体，衣服之外，孩子得到的是一条可以一直往上走的成长旅程。

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

盘点电商行业资讯，时刻了解行业政策、融资动态、扫描电商热点事件。欢迎留言评论区讨论。 国内资讯 【国内电商】 万物新生集团二季度收入66.1亿元 同比增长32.4% 8月23日消息，二手消费电子产品交易和服务平台万物新生（爱回收）集团近日发布的2026年第二季度业绩报告显示，报告期内集团总收入66.1亿元，同比增长32.4%，超出收入指引区间高端。其中1P（自营）产品销售收入61.9亿元，3P（平台）服务收入4.1亿元。 东方甄选2026财年营收57亿元 归母净利润大涨9378% 8月22日消息，东方甄选发布2026财年全年业绩公告。财报显示，2025年6月1日至2026年5月31日，东方甄选实现营收57.01亿元，同比增长29.8%；归母净利润5.44亿元，同比增长9377.8%。若剔除“与辉同行”直播间带来的营收影响，公司营收同比增长36.3%；毛利为20亿元，同比增加45.2%；经营利润为6.6亿元，相比去年同期实现扭亏为盈。全财年整体GMV达到102亿元，其中自营产品GMV达54亿元，占总GMV比重52.6%。截至报告期末，自营SPU数量从732个扩充至1009个。 京东联合四省六地发起国内首届“京东开羊季” 8月21日消息，京东近日联合内蒙古呼伦贝尔新巴尔虎右旗、锡林郭勒盟，宁夏盐池，新疆和田、阿勒泰，以及甘肃临夏州等四省六地核心畜牧产区，共同发起国内首届“京东开羊季”，助力牧区特色畜牧产品上行、农牧民稳定增收。8月20日，“京东开羊季”首站启动仪式在呼伦贝尔新巴尔虎右旗举行。京东与当地市场监管局现场签署战略合作协议，双方将建立政企协同合作机制，共同维护“西旗羊肉”地标品牌形象和市场秩序。 阿里最新财报：速卖通实现盈利 8月20日，阿里发布2026年6月季度财报。财报显示，国际电商业务方面，得益于物流优化和成本效益提升，速卖通本季度已在经营层面实现盈利，本地商品销售额占比同比持续显著提升。 苏宁易购与苏豪控股集团达成合作 开拓中亚市场 8月21日消息，中亚进出口博览会暨2026江苏进出口商品展览会在哈萨克斯坦阿拉木图举办。展会首日，苏宁易购与苏豪控股集团正式签署《中亚江苏中心入驻合作协议》，达成深度战略合作。双方将以双向进出口、内外贸联动为方向，以产品定制供应链为依托，联合建设全链路电商销售平台，配套完善跨境物流、资金结算体系，同时联合本地合作伙伴推进本地化运营能力建设。 阿里吴泳铭：平头哥AI芯片下半年将持续放量 8月21日消息，阿里巴巴集团日前发布2027财年第一季度财报。在分析师电话会上，集团CEO吴泳铭表示，平头哥已建立覆盖GPU、CPU、网络芯片的全栈自研芯片组合。真武芯片到8月初已经服务了超过650家客户。基于新一代平头哥芯片真武M890的超节点实例，近日已上线阿里云进行规模化销售，下半年将持续放量以满足客户的旺盛需求。 阿里蒋凡：即时零售板块预计2029财年实现整体盈利 8月21日消息，阿里巴巴阿日前发布2027财年第一财季财报。电话会上，阿里中国电商事业群CEO蒋凡表示，“我们认为非餐饮品类的即时零售交易额规模将在下个财年内超过餐饮品类，会带动电商大盘很多品类，即时零售板块预计在2029财年实现整体盈利，有望成为电商板块的第二曲线。”接下来会加速整合盒马和天猫超市的相关板块，发展非餐饮品类的即时零售，尤其是加速发展前置仓。 吴泳铭：阿里AI年化收入突破495亿元 8月21日消息，阿里巴巴集团日前发布2027财年第一季度财报。在分析师电话会上，阿里巴巴集团CEO吴泳铭表示，AI已成为阿里云加速增长的核心引擎。本季度，阿里AI相关产品年化收入（ARR）突破495亿元人民币（73亿美元），在阿里云外部商业化收入中的占比升至35%。AI相关产品毛利率显著高于云产品平均水平。预计下季度AI相关产品ARR将接近100亿美元。 阿里2027财年Q1资本支出为676.78亿元 同比增长75% 8月20日消息，阿里巴巴公告，截至6月30日结束的季度内，资本支出为676.78亿元，较上年同期的386.76亿元人民币增长75%，源于持续投入AI基础建设，以满足强劲且不断增长的客户需求。同比增长显著，受多项因素驱动，包括采购周期波动、预期客户持续采用AI智能体带动的CPU算力增加，以及各类芯片组件价格上涨。 阿里巴巴一季度营收2690亿元 同比增长9% 8月20日消息，阿里巴巴发布2027财年第一财季财报。一季度，阿里巴巴营收2690亿元，同比增长9%；非公认会计准则下，净利润为207.15亿元。其中，阿里云外部商业化收入加速增长45%，增速创22个季度新高；经调整EBITA同比增长133%，EBITA利润率增至12%；AI相关产品收入连续第十二个季度实现三位数同比增长，AI已成为云增长的确定性引擎。大消费方面，即时零售收入同比增长45%，CMR（客户管理收入）同比增长1%；淘宝闪购市场份额稳固，UE环比持续改善。 拼多多最新财报：研发同比增长40%，平台生态治理成效显现 8月24日消息，拼多多发布最新财报，二季度营收1124亿元，Non-GAAP研发投入43亿元，同比增长40%。今年以来，拼多多持续加码对产业发展、技术研发和平台治理的投入力度。在平台治理上，拼多多持续加码资源投入，提升技术防控体系，扩充专业治理团队，将治理触角延伸至全域的细分品类，仅6月份就集中落地50余个细分领域治理方案，为广大商家和消费者打造了一个更安全、更规范、更值得信赖的生态环境。 拼多多二季度营收1124亿元 同比增长8% 8月24日消息，拼多多发布2026年二季度财报。财报数据显示，二季度，拼多多总收入为人民币1124亿元，比2025年同期人民币1040亿元增长8%。拼多多本季度营业利润为人民币278亿元，同比增长8%；Non-GAAP营业利润为人民币291亿元，同比增长5%。本季度归属于普通股股东的净利润为人民币272亿元，同比下降12%；归属于普通股股东的Non-GAAP净利润为人民币285亿元，同比下降13%。 淘宝发布「IFASHION黑标」：仅17家中国原创服饰品牌入选 日前，淘宝正式发布「IFASHION黑标」。百万服饰商家中，仅17家原创品牌入选首批。这17席要连过三关：设计、面料、口碑的多维硬指标筛选，消费者"购物车投票"，以及专业黑标理事会的综合验证。入选品牌被视为线上服饰商家品牌化的最高水平，也是消费者可以"闭眼入"的新一代中国顶级原创服饰品牌。 闲鱼举办跨行业卖家大会：无忧卖上线4个月服务300万卖家，六成商品清仓提速5倍 近日，闲鱼跨行业卖家大会在杭州、深圳、上海三地相继举办，邀约来自20个行业的300余位卖家，共同探讨规范经营与生意增长。大会围绕合规经营与无忧卖两大主题，设置合规经营解读、行业卖家运营成功案例分享、现场答疑与账号官方1V1诊断等议程。会上，闲鱼披露了无忧卖最新数据：上线4个多月以来，无忧卖已吸引近300万卖家加入，无忧卖商品流量及订单量大幅增长，近六成商品清仓速度较日常提升至5倍。 拼多多赵佳臻：推动平台生态治理常态化 精细化和制度化 8月24日消息，8月24日，拼多多发布截至6月30日的2026年第二季度财报。受益于“千亿扶持”等惠商战略的效用释放，以及平台治理体系的全面升级，拼多多二季度营收达到1124亿元，利润272亿元。其中，Non-GAAP研发投入43亿元，同比增长40%。 配售800亿后，蔡崇信 吴泳铭再出手增持1.2亿港元阿里股票 8月24日消息，据港交所最新信息，在阿里巴巴宣布800亿港元新股配售计划后，集团主席蔡崇信、CEO吴泳铭出手增持合计约1.2亿港元阿里股票。据了解，蔡崇信此次买入72万股阿里巴巴港股，均价约112港元，耗资约8000万港元；CEO吴泳铭买入35万股阿里巴巴港股，均价约111.6港元，耗资约4000万港元。两人合计增持107万股，总金额约1.2亿港元。 闲鱼举办跨行业卖家大会：无忧卖上线4个月服务300万卖家 六成商品清仓提速5倍 8月24日消息，闲鱼跨行业卖家大会在杭州、深圳、上海三地相继举办，邀约来自20个行业的300余位卖家，共同探讨规范经营与生意增长。大会围绕合规经营与无忧卖两大主题，设置合规经营解读、行业卖家运营成功案例分享、现场答疑与账号官方1V1诊断等议程。会上，闲鱼披露了无忧卖最新数据：上线4个多月以来，无忧卖已吸引近300万卖家加入，无忧卖商品流量及订单量大幅增长，近六成商品清仓速度较日常提升至5倍。 苏宁1元挂牌转让云晟电商 132亿负债剥离B2C旧资产 8月24日消息，苏宁易购近日通过董事会议案，拟以1元象征性价格挂牌转让全资子公司江苏云晟电子商务有限公司100%股权。该公司原为苏宁易购电商业务总部，目前负债规模达132亿元，净资产为-102亿元，此次交易实质上是对早期B2C电商核心资产进行整体剥离，实现相关资产出表。曾投入大量资源建设线上商城、收购PPTV以及家乐福中国的苏宁易购，近年来在平台化竞争与即时零售发展双重压力下逐渐承压。为缓解经营压力，公司陆续通过关闭亏损门店、出售家乐福股权以及盘活物流资产等方式调整业务结构。此次以1元转让云晟电商，被视为苏宁进一步隔离历史亏损资产、减少旧电商业务对上市公司财务表现影响的重要举措。 闪送二季度完成订单6310万单 环比增长9% 8月24日消息，闪送近日发布2026年第二季度业绩报告。财报数据显示，2026年第二季度，闪送实现营收9.4亿元。受市场竞争加剧影响营收有所回调，但平台多项核心经营指标保持稳健，二季度毛利润为9550万元，经调整净利润1140万元，商业模式内生价值持续释放。订单规模保持稳定体量，二季度完成订单6310万单，环比增长9%。 【直播电商】 韦东奕B站账号更新 带货橱窗上架39.8元数学练习册 8月21日消息，北京大学教师韦东奕B站账号时隔一年多再次更新动态，发布一条标题名为“一道有趣的题”的视频，其中列举了数道数学题目并给出解法。同时，该账号的“小店”带货橱窗上新了一本小学数学解题能力训练的练习册，售价39.8元，作者为韦东奕。韦东奕亲属回应：该账号是韦东奕本人的，视频也是他本人发布的。另一亲属表示：（账号）不是冒充的，书也是韦东奕自己主编的。 【快递】 顺丰同城“骑士年度激励计划”已覆盖全国28城 8月23日消息，继今年7月首批13城试点落地后，顺丰同城“骑士年度激励计划”近日迎来扩容，新增嘉兴、台州、厦门、珠海、青岛、济南、石家庄等15座城市，同步追加投入近600万元，累计投入达1200万元。作为即时配送行业内首个面向骑士推出的“年金”式长效激励方案，顺丰同城“骑士年度激励计划”通过制度化、累积式的年度激励账户，为长期服务的骑士提供与之匹配的稳定回报。此次扩容后，该计划覆盖全国28座城市，涵盖一线、新一线、区域中心城市及地级市，将更多区域的一线配送骑士纳入长效保障体系。 菜鸟CEO万霖：将在多个APEC经济体建设机器人仓储 8月22日消息，2026年亚太经合组织（APEC）海关与商界对话会在大连举行，菜鸟集团CEO万霖作为中方企业代表在会议中发言表示，跨境电商物流已成为覆盖广、效率高、科技新的全球产业基础设施，菜鸟将继续通过AI等智能技术提升跨境物流效率、合规水平和绿色能力，助力APEC各经济体电商高质量发展，更好实现“全球买、全球运”。万霖透露，菜鸟未来将在多个APEC经济体投资建设机器人仓储设施。 京东宠物专送服务“京宠达”覆盖范围扩充至45城 8月22日消息，京东物流宣布旗下宠物专送服务“京宠达”再度扩容，覆盖新增长沙、扬州、泰州等城市。“京宠达”宠物专送服务于今年6月1日正式上线，服务聚焦猫、狗两大类主流伴侣宠物，为宠物商家提供一站式门到门专送解决方案。最新一轮扩容后，“京宠达”始发城市涵盖沈阳、鞍山、辽阳3城，可配送范围覆盖北上广深杭等45城，均可实现“今日下单、明日送达”。 跨越速运与广州白云机场达成航空物流合作 8月22日消息，跨越速运与广东省机场集团物流有限公司日前签署航空物流合作协议，双方将基于广州白云机场空侧场地展开深度合作。这是继2022年华南航空枢纽空侧在深圳宝安国际机场启用后，跨越速运在华南区域航空物流版图上落下的又一关键棋子。从深圳单核支撑到广深双核运转，集团华南航空资源迎来质的升级，也为即将到来的货运旺季筑牢坚实保障。 极兔速递佛山转运中心正式启动 8月22日消息，极兔速递佛山转运中心近日正式启动。该项目总投资约1.2亿元，于今年8月全面投入运营，建成后将进一步强化佛山及周边区域的快件集散、中转和运输能力。目前，中心可实现全天候连续作业，日均快件操作能力约220万票，单日峰值处理能力可达330万票。随着中心正式投运，将有效提升佛山及粤西片区在业务高峰期的快件承载能力，进一步增强珠三角西部区域运行的稳定性和韧性。 极兔速递佛山转运中心正式启动 8月22日消息，极兔速递佛山转运中心近日正式启动。该项目总投资约1.2亿元，于今年8月全面投入运营，建成后将进一步强化佛山及周边区域的快件集散、中转和运输能力。目前，中心可实现全天候连续作业，日均快件操作能力约220万票，单日峰值处理能力可达330万票。随着中心正式投运，将有效提升佛山及粤西片区在业务高峰期的快件承载能力，进一步增强珠三角西部区域运行的稳定性和韧性。 顺丰7月合计营收263.68亿元同比增长6.12% 8月21日消息，四家快递企业2026年7月业绩数据显示，行业呈现“顺丰提价、通达系扩量”的差异化发展态势。顺丰控股实现合计营收263.68亿元，同比增长6.12%。其中，速运业务营收191.47亿元，同比增长2.63%，业务量为13.25亿票，同比下降3.78%，单票收入提升至14.45元，同比增长6.64%；供应链及国际业务实现营收72.21亿元，同比增长16.66%，进一步体现其通过控制业务规模、提升单票价值以及国际业务补充增长的策略效果。圆通实现营收59.67亿元，同比增长11.10%，业务量达到28.95亿票，同比增长12.07%，单票收入为2.06元，同比小幅下降0.86%，整体保持规模与效率平衡。 顺丰同城“骑士年度激励计划”已覆盖全国28城 8月24日消息，继今年7月首批13城试点落地后，顺丰同城“骑士年度激励计划”近日迎来扩容，新增嘉兴、台州、厦门、珠海、青岛、济南、石家庄等15座城市，同步追加投入近600万元，累计投入达1200万元。作为即时配送行业内首个面向骑士推出的“年金”式长效激励方案，顺丰同城“骑士年度激励计划”通过制度化、累积式的年度激励账户，为长期服务的骑士提供与之匹配的稳定回报。此次扩容后，该计划覆盖全国28座城市，涵盖一线、新一线、区域中心城市及地级市，将更多区域的一线配送骑士纳入长效保障体系。 【其他】 闪送二季度完成订单6310万单 环比增长9% 8月23日消息，闪送近日发布2026年第二季度业绩报告。财报数据显示，2026年第二季度，闪送实现营收9.4亿元。受市场竞争加剧影响营收有所回调，但平台多项核心经营指标保持稳健，二季度毛利润为9550万元，经调整净利润1140万元，商业模式内生价值持续释放。订单规模保持稳定体量，二季度完成订单6310万单，环比增长9%。 淘宝闪购联合盒马等连锁商超开启“全国开海季” 8月22日消息，全国开海季正式拉开大幕，淘宝闪购联合盒马、永辉超市、三江购物、乐家超惠等连锁商超以及菜老包生鲜菜场连锁上架开售海产，淘宝闪购开渔近一周，梭子蟹订单量环比上涨260%，其他水产品也呈现上涨态势。三眼蟹、红花蟹、兰花蟹等鲜活蟹类，以及带鱼、鲳鱼、鱿鱼、龙头鱼等冰鲜海鱼也将随着各地开海陆续开售。 淘宝闪购：未来3年投入10亿元建设校园服务 8月21日消息，淘宝闪购2026高校生态峰会日前在杭州举行。会上，淘宝闪购发布《“1+N”多元业态白皮书》，并正式启动“登云工程”：计划未来3年投入10亿元规模资源，全力推动全国校园信息化建设和产业升级。同时，淘宝闪购还与上海交通大学、东北大学、西南大学等多所高校签署战略合作协议，共同探索校园服务数字化建设。 淘宝闪购“蜂鸟跑腿”品牌升级为“淘宝闪购跑腿” 8月21日消息，淘宝闪购近日将“蜂鸟跑腿”品牌正式升级为“淘宝闪购跑腿”。在手机淘宝相关频道，该服务入口名称现已同步更新为“淘宝闪购跑腿”。公开资料显示，蜂鸟跑腿服务规模位居行业第一梯队，帮买、帮送、帮忙等服务7×24小时运营，覆盖全国500多个城市。服务场景涵盖帮取餐品，取送文件及快递，蛋糕礼品帮买帮送等。 美团外卖上线“商场店”“商业街店”等标签 8月21日消息，美团外卖近日已在部分商家页面增加“商场店”“商业街店”“平台巡检过”等经营标签。截至今年8月，美团上已有超20万家商家展示“商场店”标签，4.5万家商家展示“商业街店”标签。新标签主要用于商场档口、商业街柜台及使用公共就餐区等难以简单归入传统“堂食店”或“无堂食店”的经营场景，帮助消费者在下单前了解门店位置、经营环境和就餐条件。 2026 胡润中国品牌榜发布：抖音、微信并列最具价值中国民营品牌 8月24日消息，胡润研究院今日发布《2026胡润中国品牌榜》（Hurun Most Valuable China Brands 2026）。在中国品牌前十名中，抖音与微信均以3000亿元品牌价值并列最具价值中国民营品牌。华为回归前十，位列中国品牌第七名（总榜第九名）。拼多多上升一位。五粮液虽下降47%至905亿元，仍位列中国品牌第九名。 京东外卖“宝藏小馆”全国上线 8月24日消息，京东外卖“宝藏小馆”全国上线，首批覆盖350城、8万多家烟火小店。 国外资讯 【国外】 南航物流开通中国至挪威全货机航线 8月23日消息，南航物流一架B777F全货机，于8月20日满载近百吨挪威冰鲜三文鱼，平稳降落在上海浦东国际机场，标志着南航物流开通上海浦东—阿姆斯特丹—奥斯陆—上海浦东航线。该航线不仅是南航开通的首条前往挪威的航线，更是其完善欧洲航空物流网络、深耕北欧市场的重要举措。该航线将以挪威三文鱼等高品质生鲜为核心货源，着力打造一条常态化的生鲜进口空中通道，在丰富国内消费供给的同时，为中挪及中国与北欧地区的经贸往来注入新动能。 怡亚通物流马来西亚公司揭牌运营 8月22日消息，怡亚通物流马来西亚公司揭牌仪式日前举行。怡亚通物流（马来西亚）有限公司坐落于马来西亚核心工业与港口枢纽——槟城。依托怡亚通集团全球供应链网络与管理体系，公司专注于马来西亚本地及中马跨境物流业务，打造集运输、仓储、分销、清关于一体的综合供应链服务平台，致力于为客户提供端到端供应链解决方案，助力企业提升交付效率、降低运营成本、增强市场竞争力。 TikTok Shop 马来西亚年内投入2.5亿林吉特扶持卖家 8月22日消息，TikTok Shop马来西亚表示，平台今年将累计投入2.5亿林吉特，用于支持马来西亚本土卖家和内容创作者。这笔投入包括通过新推出的JomLokal Booster 2.0活动追加7000万林吉特，根据计划，本地企业将获得物流、推广活动等方面的支持，平台将零平台佣金范围扩大至另外44个产品子类别。消费者也能获得每笔订单2林吉特的标准配送费返还，并可使用海运配送至东马地区以及覆盖全国的自提点。 沃尔玛Q2营收1879亿美元 全球电商销售额增长23% 8月21日消息，沃尔玛公布截至2026年7月31日的2027财年第二季度业绩。本季度，沃尔玛营收营收为1879亿美元，同比增长5.9%；净利润为63.7亿美元，较上年同期的70.3亿美元有所回落；营业利润约93亿美元，同比增长28.8%；按固定汇率计算，调整后营业利润增长17.4%。在核心业务增长方面，全球电商销售额增长23%，主要由门店履约的提货和配送以及Marketplace业务推动。 菜鸟旗下递四方上线TikTok Shop欧盟12国直邮产品 8月21日消息，菜鸟集团旗下递四方近日推出面向TikTok Shop欧盟12国的跨境直邮3PL产品，覆盖德国、法国、意大利、西班牙、奥地利、比利时、荷兰、波兰、捷克、匈牙利、葡萄牙、希腊12个核心欧盟站点国家，主打8–12天稳定送达、99%妥投率及运费至高六折，为跨境卖家提供兼顾时效、合规与成本的一站式履约服务。该产品覆盖的12国与TikTok Shop欧洲欧盟站点完全对齐，商家一站接入即可实现全欧盟发货，无需分头对接多家服务商。 亚马逊扩大与印度铁路合作 启动西部专用货运走廊运输 8月21日消息，亚马逊印度公司已扩大与印度铁路的合作，开始在西部专用货运走廊（WDFC）上运营，在节日季前夕将其铁路物流网络扩展至印度北部和西部地区。这项新增的铁路运输服务将采用超10个Van Parcel货运车厢，每天运行，连接亚马逊位于德里、艾哈迈达巴德和苏拉特的物流网络，并进一步衔接浦那、孟买和果阿等市场。 大重货跨境机会升温，eBay持续加码大重货出海 海外市场对大件、重货商品的需求持续释放，大重货跨境业务正成为跨境电商新的增长点。依托中国制造的产业基础和供应链优势，高客单价、专业需求驱动的大重货，也为中国卖家拓展海外市场、寻找新增量提供了更多机遇。顺应这一趋势，eBay持续加大面向中国卖家的大重货业务投入。近日，eBay 2026年第三季度（Q3）美国站大重货促销计划进一步扩大激励覆盖范围。 Shopee 菲律宾8月31日起下线买家端“店铺消息”入口 8月23日消息，Shopee菲律宾本土店发布客服入口调整通知，自2026年8月31日起，买家端Shopee App中Chat页面的“店铺消息（Shop Message）”板块将正式下线，卖家无法再通过买家端App处理咨询消息。后续卖家仅可通过Seller Center App中的Chat标签页，或卖家中心网页版Webchat（seller.shopee.ph/new-webchat/conversations）回复买家。9月15日，全新版Shopee Seller Center App将在App Store和Google Play上线。 TikTok Shop 欧洲跨境POP发布“领航计划” 投入10亿级资源激励 8月21日消息，TikTok Shop欧洲跨境POP业务正式发布“领航计划”，宣布投入10亿级资源激励，帮助中国商家以更低门槛、更高效率打开英欧市场。 沃尔玛Q2电商占美国整体业务23% 关税退款推动毛利率提升 8月21日消息，日前，沃尔玛发布了 2027 财年第二季度财报。沃尔玛实现总营收1864亿美元，同比增5.1%；调整后营业利润92亿美元，同比增17.4%；调整后每股收益0.81美元，同比增19.1%。财报发布后，沃尔玛股价盘前跌超10%。 欧盟第20轮对俄制裁拉黑6家中企 扬杰电子一电航空等涉军民两用物资 8月21日消息，欧盟4月23日通过第20轮对俄制裁包，在新增黑名单中一口气列入6家中国企业——香港Brightmile Ltd、扬州扬杰电子、上海ETS Solutions、湖南浩天亿、北京西超国际科技贸易、深圳一电航空技术，制裁代码Entry 1987—1992，理由统一为“向俄罗斯军工综合体供应军民两用物资”。中国商务部随即强硬回应，重申反对未经联合国授权的单边制裁与长臂管辖，要求欧方立即移出清单并警告“中方将采取必要措施，一切后果由欧方承担”。 Wildberries8 月31日起FBS超48小时将按小时收取罚款 8月21日消息，Wildberries宣布自2026年8月31日起对FBS（卖家自发货）模式实施发货规则调整，延迟发货的“额外佣金加价”机制将正式改为按小时计算的卢布罚款。新规下，13小时内完成发货可获得5个百分点佣金减免，13—42小时发货可减免3.5个百分点，42—48小时维持基础佣金水平；超过48小时后，将从第48小时起按照小时阶梯递增罚款，其中48—54小时加罚0.3个百分点，54—60小时加罚0.35个百分点，60小时以上加罚0.45个百分点。此次罚款机制将覆盖已购买订单以及被买家取消的订单，旨在避免卖家通过等待订单取消来规避发货责任。 SHEIN 布局双节户外花园品类 万圣节与圣诞节主题商品成上新重点 8月21日消息，SHEIN近期发布户外花园用品万圣节与圣诞节双节点招品指南，要求具备主题货盘的商家在黑五网一前完成商品提报与上新。万圣节方向重点布局门廊、草坪、花园小径等场景，打造“搞怪派对庭院布景”，核心色彩包括红、紫、黑、白、橙，主要元素涵盖南瓜、骷髅、黑猫、蜘蛛网等，相关品类聚焦园艺装饰、摆件、野餐垫以及户外地垫。圣诞节方向则围绕红绿主色调展开，融入圣诞老人、驯鹿、雪花、姜饼、松枝等节日元素，将户外地毯、桌布椅罩、花盆、花园地插、装饰石头等产品纳入庭院聚会场景。 Similarweb 7 月全球电商流量榜：Amazon以27.25亿访问量居首 Temu位列第二 8月21日消息，Similarweb最新7月全球电商流量榜显示，Amazon以27.25亿次月访问量稳居首位，并形成明显领先优势；Temu以11.10亿次访问量排名第二，环比下降3.16%。eBay、Ozon、Walmart、AliExpress等平台的月访问量则集中在5亿—7亿次区间，形成第三梯队，全球电商流量格局呈现“一超一强+多极跟随”的分层特征。Temu访问量环比小幅回落，反映出低价补贴逐步退潮以及多市场合规成本上升背景下，自然流量进入再平衡阶段。与此同时，Amazon近27亿次的月访问量进一步体现出其搜索货架与Prime履约体系在购物流量领域的优势。 近半数卖家撤离平台仓 俄罗斯电商履约模式转向自主控货 8月21日消息，俄罗斯电商平台的集中式仓储模式正在发生变化。2026年上半年，约45%—50%的卖家已将部分或全部库存从平台仓（FBO）转移至自有仓库。此前占据约60%市场份额的FBO模式，因平台流量排名优势减弱以及仓储成本持续上涨，逐渐面临调整。DPD电商平台负责人Klim Krapp表示，市场对替代物流方案的需求明显增加，仅过去两周，DPD的DBS物流需求增长12倍。同时，100—500平方米的小型配送仓库需求也持续提升。越来越多卖家倾向采用“小仓分散+多渠道分流”的方式替代依赖大型平台仓，以便在平台限流、扣货或旺季仓储压力增加时，能够快速调整库存并切换至其他销售渠道。 亚马逊扩大与印度铁路合作 启动西部专用货运走廊运输 8月21日消息，亚马逊印度公司已扩大与印度铁路的合作，开始在西部专用货运走廊（WDFC）上运营，在节日季前夕将其铁路物流网络扩展至印度北部和西部地区。这项新增的铁路运输服务将采用超10个Van Parcel货运车厢，每天运行，连接亚马逊位于德里、艾哈迈达巴德和苏拉特的物流网络，并进一步衔接浦那、孟买和果阿等市场。 阿里最新财报：速卖通实现盈利 Brand+品牌买家渗透超30% 8月21日消息，日前，阿里发布2026年6月季度财报。财报显示，国际电商业务方面，得益于物流优化和成本效益提升，速卖通本季度已在经营层面实现盈利，本地商品销售额占比同比持续显著提升。 Shopee 越南站卖家利好：后台一键绑定Payoneer 资金管理更高效 8月21日消息，Payoneer派安盈目前已与Shopee平台实现API对接。Shopee越南站3PF卖家只需在Shopee后台授权绑定Payoneer账户，平台发起打款后，即可直接接收平台回款，并支持越南盾原币入账。该方案进一步简化了入账流程，有助于提升卖家的资金流转和运营效率。此外，即日起至2026年10月31日，符合条件的Shopee越南站3PF店铺可报名参与提现手续费返现活动。如需报名或了解活动详情，可在Payoneer官方渠道联系查询。 亚马逊Prime Video拟在拉美投资20亿美元 8月21日消息，亚马逊旗下Prime Video宣布，将在2027年至2030年期间向拉丁美洲投资20亿美元，该投资将涵盖墨西哥、巴西、阿根廷、哥伦比亚和智利等国的原创节目、本地制作及引进内容，以及体育赛事直播权。 Shopee 新加坡本土店推0.5%佣金返还 加入AdXpress至年底可领 8月21日消息，Shopee新加坡本土店宣布，为鼓励卖家进一步使用平台广告工具，符合条件的受邀卖家在参与AdXpress广告订阅计划期间，可获得0.5%的基本佣金返还。该返还政策持续至2026年12月31日或卖家退出项目，以较早发生的时间为准。AdXpress本质上是一项广告管理工具，可按照销售额固定比例自动划转为Shopee广告积分，用于持续投放，无需卖家手动充值。本次返佣主要针对“广告投资率低于市场均值4%且未持续参与”的卖家，已注册旧版AdXpress的卖家不在邀请范围内。平台通过“返佣+自动充值”机制，进一步提升中小卖家的广告参与率。 亚马逊成为西部专用货运走廊首家电商租户 印度铁路网络扩展至100多个城市 8月21日消息，亚马逊近日宣布扩大与印度铁路公司（Indian Railways）及印度专用货运走廊公司（DFCCIL）的合作，成为首家在西部专用货运走廊（Western DFC）首列JPP-RCS列车上运营的电商企业。此次新增铁路服务采用超过10个Van Parcel货运车厢，每日运行，覆盖德里、艾哈迈达巴德、苏拉特，并连接浦那、孟买和果阿等地区，使亚马逊铁路运输网络从2019年的1列扩展至140多列，覆盖130多条城际线路及100多个城市，为即将到来的印度节日购物季提前做好运力储备。该布局是在2019年首次使用铁路快件、2023年进入专用货运走廊基础上的持续推进。 法国电商2025年逆势增长7%至1964亿欧元 AI成为94%企业标配工具 8月21日消息，法国电商与远程销售联合会（Fevad）发布的最新《2026年电商关键数据》显示，2025年法国电商市场实现逆势增长，全年营业额达到1964亿欧元，同比增长7%，占零售总额的12%。其中，商品销售业务恢复增长，同比增长4%；服务业务表现更为突出，同比增长9%。全年线上交易数量达到32亿笔，同比增长11%，共有4220万名网购用户，占16—74岁人口的80%。消费者年均线上消费金额为4657欧元，平均下单75次，智能手机购物和多元化支付方式逐渐成为主流。法国电商行业整体规模持续扩大，目前活跃电商网站数量超过15.8万个，同比增长7%；相关就业岗位达到23.4万个，同比增长9%。 Similarweb 7 月全球电商流量榜：Amazon以27.25亿访问量居首 Temu位列第二 8月21日消息，Similarweb最新7月全球电商流量榜显示，Amazon以27.25亿次月访问量稳居首位，并形成明显领先优势；Temu以11.10亿次访问量排名第二，环比下降3.16%。eBay、Ozon、Walmart、AliExpress等平台的月访问量则集中在5亿—7亿次区间，形成第三梯队，全球电商流量格局呈现“一超一强+多极跟随”的分层特征。Temu访问量环比小幅回落，反映出低价补贴逐步退潮以及多市场合规成本上升背景下，自然流量进入再平衡阶段。与此同时，Amazon近27亿次的月访问量进一步体现出其搜索货架与Prime履约体系在购物流量领域的优势。 近半数卖家撤离平台仓 俄罗斯电商履约模式转向自主控货 8月21日消息，俄罗斯电商平台的集中式仓储模式正在发生变化。2026年上半年，约45%—50%的卖家已将部分或全部库存从平台仓（FBO）转移至自有仓库。此前占据约60%市场份额的FBO模式，因平台流量排名优势减弱以及仓储成本持续上涨，逐渐面临调整。DPD电商平台负责人Klim Krapp表示，市场对替代物流方案的需求明显增加，仅过去两周，DPD的DBS物流需求增长12倍。同时，100—500平方米的小型配送仓库需求也持续提升。越来越多卖家倾向采用“小仓分散+多渠道分流”的方式替代依赖大型平台仓，以便在平台限流、扣货或旺季仓储压力增加时，能够快速调整库存并切换至其他销售渠道。 沃尔玛将推出全新女装品牌Scenario 8月24日消息，沃尔玛正在进一步加码服装业务。市场消息称，沃尔玛本周将在美国推出全新女性服装品牌Scenario，主要面向更年轻、更关注时尚趋势的女性消费者。品牌首批将推出数百款商品，包括刺绣衬衫、褶裥牛仔衬衫、人造皮革包等，大多数商品售价低于25美元（约合人民币168元）。 纯新增量！速卖通用AI为商家一口气开200多个新市场 AI自动帮跨境商家拿订单的时代来啦！ 嘉友国际首个非洲海外仓正式开工建设 8月24日消息，嘉友国际发文称，8月22日，公司纳米比亚鲸湾海外仓项目正式开工建设。项目占地18公顷，是公司完善中南部非洲跨境物流网络、非洲战略落地的重要里程碑。嘉友国际表示，该项目建成投用后，将与公司刚果（金）、赞比亚陆路口岸项目、跨境运输车队实现高效协同，逐步构建起“跨境陆运+口岸枢纽+港口仓储”一体化跨境综合物流服务体系。 TikTok Shop 新加坡GMV同比大增1.8倍 多站点运营工具及“马发新”服务上线 8月24日消息，最新数据显示，2026年上半年，TikTok Shop新加坡站点GMV达去年同期1.8倍以上，接近翻番。其中，直播场域表现尤为亮眼，GMV同比涨96%；平台购买用户数快速攀升，达去年同期1.6倍。2026年6.6大促期间，新加坡站点多场域协同发力，日均GMV创下历史新高。 宝洁两年在俄发起288起商标诉讼 Wildberries和Ozon假货卖家面临处罚 8月24日消息，宝洁于2025年1月至2026年8月19日期间，针对俄罗斯电商卖家发起288起商标侵权诉讼，其中2026年已涉及90起。相关诉讼主要针对Wildberries、Ozon平台上销售假冒汰渍、Lenor、Fairy、潘婷、海飞丝等品牌商品的商家，部分案件还将平台列为第三方共同诉讼对象。目前，俄罗斯法院已确认多起假冒商品案件并作出处罚。例如，喀山Zebra Opt LLC因未经授权使用Tide、Pantene等商标，被判处罚款5万卢布；个体经营者比谢罗娃、克里莫娃则分别通过庭外和解支付6.3万卢布和5.4万卢布，并承诺再次违规将承担更高处罚。 eBay 欧洲加速计划上线 助力英国卖家一键拓展德法意西市场 8月24日消息，eBay近日推出全新的“欧洲加速计划”（EAP），取代此前的“欧洲销售助推器”，该计划面向具备“特色店铺”或“旗舰店铺”资格且信誉良好的英国卖家开放。依托Webinterpret技术，卖家可自动完成商品信息翻译、货币转换以及库存同步，实现商品Listing本地化处理。单个店铺最多可将5万条Listing（每个国家站点上限1万条）同步至德国、法国、意大利、西班牙四大欧盟市场，同时需满足向目标国家提供直邮服务的要求。该计划通过后台功能整合，将跨语言、跨币种以及跨物流环节的运营门槛进一步降低，英国卖家无需额外设立欧盟主体即可触达四国消费市场。 亚马逊8月24日更新BSA第18条 店铺回款权转让质押需获平台书面批准 8月24日消息，亚马逊自2026年8月24日起更新《商业解决方案协议》（BSA）第18条，明确规定未经平台事先书面同意，卖家不得向第三方转让或质押协议项下的任何权利与义务，包括销售收益收取权。该调整直接限制了利用未来店铺回款进行融资、将应收款转让给保理商以及私下变更店铺主体等操作。此次更新将旧版“不得转让入驻协议”的限制范围进一步扩大，覆盖经营权利与履约义务的整体绑定。平台同时保留对原始法人及受让方的长期追责权，违规未处理、TRO（临时限制令）未解决、营业执照存在异常等问题店铺的转让渠道也将受到限制。对于跨境卖家而言，店铺将从“可交易资产”重新回归为与经营主体责任绑定的业务载体。 TikTok 美区圣诞礼卡书7天售出7045单 34秒视频带动3636件销量 8月24日消息，一款尺寸为5×7英寸、售价12—13美元的圣诞礼品卡书近日登上TikTok美区电脑办公类目销量榜首。该产品采用螺旋装订设计，支持消费者DIY放入礼品卡、现金或手写祝福。店铺Rddmo Direct在近7天（8.15—8.21）内售出7045单，GMV超过8.45万美元。 俄边境投资170亿卢布建设超级跨境物流中心 8月24日消息，俄罗斯后贝加尔斯克口岸附近将建设一座投资170亿卢布（约17亿元人民币）的超级跨境物流中心。后贝加尔边疆区政府与俄气工业银行近日签署意向协议，计划在毗邻中俄重要陆路枢纽外贝加尔斯克的区域建设该项目。该口岸承担中俄双边约60%的铁路货运业务，新物流中心建成后预计年吞吐能力达到1500万吨，并连接A-350联邦公路及中俄铁路网络，形成集智能仓储、快速通关和多品类处理于一体的综合型物流节点，覆盖食品、农产品原料、机电产品及工程机械等货物。随着后贝加尔斯克—满洲里第二铁路线持续推进，该口岸年通行能力预计将增加1100万吨，列车单向日通行能力也将实现翻倍。 Boohoo 因虚假折扣在法国被罚230万欧元 8月24日消息，英国时尚电商Boohoo因涉嫌误导性促销及产品标识违规，被法国消费者保护机构DGCCRF处以230万欧元罚款。调查发现，在Boohoo被检查的促销活动中，40%的活动并未形成真实折扣，7%的折扣幅度低于宣传标准，另有48%的促销活动实际导致商品价格隐性上涨，消费者所看到的“优惠”与商品实际价格变化存在不一致情况。此外，Boohoo还被发现将部分塑料制品标注为“皮革”或“麂皮”，违反了法国相关产品标识规定。Boohoo表示，此次调查涉及的问题主要发生于2023年10月至2024年2月期间，属于此前管理层阶段产生的问题。目前，公司已完成相关整改，并配合监管部门持续审查价格制定及产品标识流程。 速卖通9月1日将上线“潜力国家地区智能经营Agent” 可接管 200 多个国家市场 8月24日消息，阿里旗下跨境电商平台速卖通AliExpress近期上线“潜力国家地区智能经营Agent”，商家一键启用，AI即可接管200多个潜力国家地区的经营，为商家带来核心市场之外的新增量。 新加坡企业被诉通过床垫贴标“洗产地”规避美国对华关税 8月24日消息，新加坡海关8月14日发布通报，已对本土企业Healthy Living Biotech (Singapore) Pte Ltd及三名个人（28岁新加坡永居殷某、37岁中国籍薛某、54岁新加坡籍辜某）提起公诉，指控其自2024年8月起将中国生产的床垫运抵新加坡后，仅通过更换外包装、加贴“Made in Singapore”标签的方式，以新加坡原产地名义申报出口美国，涉案货值约190万新元（约1000万元人民币），涉嫌规避美国对华加征关税。 大批亚马逊卖家集中出售店铺 行业迎来结构性调整 8月24日消息，近期亚马逊卖家圈出现集中出售店铺现象，大量中小卖家赶在8月24日BSA协议第18条新规生效前挂牌转让。部分带R标的老店售价低至1万元，部分稳定盈利的私模店铺也因现金流压力选择变现退出。当前抛售卖家主要包括亏损止损、微利但受头程、广告及退货成本挤压、盈利后偿还债务以及闲置店铺规避风险四类群体。此次新规的核心变化在于，亚马逊将旧版“不得转让入驻协议”范围扩大至禁止转让或质押协议项下全部经营权利与履约义务，私下过户、回款质押以及主体代持等方式将失去操作空间。同时，存在违规未处理、TRO未解决、执照异常等问题的店铺将难以完成转让，平台仍保留对原始法人的长期追责权。 马士基黄金周暂停3个欧洲航线航次 8月24日消息，随着中国国庆黄金周临近，马士基发布远东—欧洲航线网络调整通知。由于假期前后市场需求下降以及港口操作人员减少，马士基将对AE15、AE12、AE1三项服务分别实施1个航次停航（Blank Sailing），共涉及3个航次。其中，AE15青岛港640W/645E航次将于9月28日离港，AE12宁波641W/648E航次将于10月8日离港，AE1上海641W/648E航次将于10月10日离港。马士基表示，将通过替代航线和结构性调整保障供应链稳定性与可预测性。客户仍可通过原有渠道进行订舱，但需自行确认调整后的船期安排。 海豚社——新消费新国货第一社群 沟 通、合作扫描下方二维码 如果觉得内容不错，请点击一下“在看”， 或者，把它转发分享给需要的人 文章原文

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

价格战打到第三年，基本触及每一个细分市场。2026年1至5月，77款车型被卷入这场无休止的降价竞赛，行业平均利润率被压缩至3.4%。 更严重的是同质化。各大车展展台上30多款SUV放在一起，短前悬、长车身、悬浮车顶，遮住车标几乎分不清谁是谁。设计语言的趋同折射出一个更深的困境：当电动化和智能化的技术红利被拉平，汽车行业似乎正在失去定义差异的能力。 但享界走了一条不一样的路，它带来了鸿蒙智行的第一台旅行车和第一台硬派SUV，不抄设计，不搞套娃，而且快速覆盖多个车型品类。2026年成都车展，享界带来了享界G9、享界V8、享界S9、享界S9T四款车型，已经成为鸿蒙智行旗下品类覆盖最完整的品牌。 一个成立不到三年的品牌，坚持差异化而不是跟随市场主流，享界能在高端市场站稳脚跟吗？ 个性不等于小众，好的供给激活需求 去年，享界亮出旅行车S9T，彼时行业共识是：这个品类在中国已经鲜有商业价值。 过去十年，旅行车在中国乘用车市场的份额长期低于1%。这让行业逐渐接受了旅行车叫好不叫座的宿命，旅行车和旅行车主也在中国被贴上了小众、冷门的标签。到2023年，主流品牌基本停止向中国市场投放新的旅行车产品。 但享界S9T证明了旅行车在中国并非没有受众，而是没有好的产品供给。 自媒体“车fans”此前发布的报告称，享界S9T上市后，传统豪华品牌的增换购比例达到90%以上。享界在一个被主流厂商忽视的细分市场里，用产品定义重新激活了需求。 截至2026年5月21日，享界S9与S9T累计交付突破60000台，享界S9系列用户满意度达98%。在今年年初杰兰路发布的新能源汽车品牌健康度研究中，享界的品牌档次认知已经与传统豪华品牌同列。 那些愿意为审美和驾驶乐趣多付一笔钱的人，过去大多只能在进口车型里做选择，价格贵、配置低、等待周期长。 享界选择做一台真正的好车，去回应这群用户。当需求本身没有被充分满足过时，用存量数据去否定它，是一种懒惰的判断。享界的做法却是反过来的：先找到那批愿意为审美和驾驶乐趣买单的人，然后用一台真正理解他们的车，把那些浮在水面之下的需求打捞上来。 这种产品逻辑的支撑点，在于对用户的精确判断。杰兰路此前发布的一份调研报告显示，享界的用户平均年龄33.9岁，家庭年收入82.3万元，近八成拥有本科或硕士学历。他们比BBA的典型用户年轻十岁上下，对技术的接受度更高，对“与众不同”的需求也更强烈。享界没有试图讨好所有人，它选择了一群愿意为审美和体验买单的人，然后围绕这群人构建了一整套产品和服务体系。 这批新增用户和由此积累的品牌势能，成为享界可以持续继承的资产。 2025年8月，享界G9小订开启72小时突破15000台，正式上市24小时后大定突破5000台。更具指标意义的是订单结构，G9的Ultra及Ultra+高配版本售价48.98万至55.98万元，在小订用户中占比达到90%。这意味着用户的真实成交价不是43万，而是50万起步。订单数据表明，高端用户愿意为这种不可替代性支付溢价。 华为常务董事、终端BG董事长余承东在发布会上表示，G9“比豪华更硬派，比硬派更豪华”，享界要打造的是唯一一台同时把越野能力、城市舒适性和智能化做到平衡的车。它不试图讨好所有人，而是为那些既要周末去野外、又要工作日见客户的用户，提供一个新的选项。 有数据显示，2025年，“方盒子”SUV市场销量已突破110万辆，2026年全年有望冲击150万辆。G9的成功给这个曾经非主流的市场再添了一把柴火。 连打胜仗之后，享界的下一步自然浮现：轿车和SUV都已站稳，剩下的空白市场是什么，又该怎么切入？ 2026年8月，享界V8在成都车展亮相，瞄准家用MPV市场。其车身长5335mm、轴距3250mm，采用2+2+3的7座布局，提供纯电与增程两种动力，增程版CLTC综合续航超1400公里，纯电版最高续航830km。高配车型搭载华为896线激光雷达，采用全车四激光雷达立体矩阵。 在此之前，市场上多数MPV的产品定义以第二排乘客为中心，驾驶席的体验是次要的。享界V8的做法是把驾驶席的体验权重提升到前所未有的高度，用更接近SUV的车身比例消解传统MPV的箱式造型的视觉印象，用鸿蒙座舱和智驾能力提升驾驶者的驾乘体验。 在此之前，鸿蒙智行已推出尊界V800/V680与智界V9，分别覆盖超豪华和宜商宜家MPV市场，V8补上的是家用MPV这一空白。 V8的亮相，意味着享界完成了从个人出行到家庭出行、从城市通勤到户外探索、从轿车到MPV的全场景覆盖。 回看享界的产品路径，它的“个性”与设计语言上的标新立异无关。这是一种产品哲学上的选择：不跟随同行已验证过的爆款品类，而是进入那些被行业忽视或放弃的细分市场，用合理的定价和技术能力重新定义品类标准。 比如旅行车可以不只是情怀车，硬派SUV可以不用舍弃豪华和舒适，MPV可以不是司机开的车。 在中国智能电动车行业陷入同质化竞争的当下，享界没有选择在同一个价位段里堆配置打价格战，而是用持续的产品形态创新，一步步拓展品牌的物理边界，同时用底层技术和用户体验的一致性，守住品牌的内核。 享界的高端化凭什么能成立？ 在鸿蒙智行的产品谱系中，享界承担的任务最为特殊，它要正面进入豪华品牌最稳固的销量腹地，在30万元以上的价格带与德系豪华品牌主流的SUV和轿车竞争。过去几年，中国汽车品牌的高端化尝试大多止步于配置整合和定价上探，真正能在用户心智和销量规模上同时站稳的，凤毛麟角。 北汽为这场仗准备了68年。北汽手里攒下了1340余项越野专利和近60亿公里的越野实测数据，它们构成了底盘调校、车身刚性和四驱系统的工程底座，是造一台好车的基本功。 今年一季度，北汽又追加了近20亿元资金，对享界超级工厂进行技术改造。改造后的冲压线检测精度达到0.01毫米，比头发丝还要细十倍。工厂里2000多台机器人协同作业，一分钟就能下线一台新车，每台车出厂前要经历38000多项全维度检测。 华为带来的则是另一套能力。华为在消费电子领域沉淀了二十年的产品定义方法论，同时手握芯片、操作系统、AI、云计算、通信等七个技术方向的全栈积累。 鸿蒙是这套能力的集大成者。鸿蒙的分布式架构和多设备无缝流转，在手机和平板上已经被验证过。 但让它在车内真正跑起来，难点不在操作系统本身，而在于开发流程的重构。汽车行业的产品开发以年为单位，一款车从立项到上市动辄三到四年。华为需要把消费电子行业积累的方法论，注入一个原本节奏完全不同的体系。软件团队要在硬件平台锁定之前完成绝大部分的功能验证，整车开发流程中软件和硬件的时间表被重新对齐，跨团队协作的节奏被重新设计。这套方法论在汽车行业并不多见。 智能驾驶是这套方法论落地最直观的证明。2026年4月，华为发布乾崑智驾ADS 5系统。相比上一代ADS 4采用的WEWA架构，ADS 5升级为WEWA 2.0架构，从“辅助驾驶系统”进化为“面向自动驾驶的AI智能体”。享界S9系列2026年7月完成ADS V5.0版本推送，享界G9则成为首款采用L3级架构设计、获批北京市时速120公里L3级自动驾驶道路测试牌照的车型。 这些能力的叠加，让享界在智能化上建立起领先身位。 2025年9月，华为与北汽签署深化合作协议，双方计划三年投入200亿元，围绕设计、智能辅助驾驶、智能座舱和品质安全四大领域持续推进。北汽集团董事长张建勇明确表示，北汽将“ALL in享界”。 截至2026年5月，享界累计交付突破60000台。根据北汽新能源的公告，2026年上半年，享界累计交付21018辆，同比增长119.58%。2026年第一季度，享界在鸿蒙智行内部销量占比超过30%，营收贡献超过60%。享界用约三成销量贡献了六成以上收入，高端化正在成为北汽新能源经营质量改善的重要支撑。 对汽车这种长决策周期的消费品来说，NPS的含金量往往高于短期销量。享界S9T还在杰兰路发布的2025年度下半年新能源汽车品牌健康度研究中，斩获中大型轿车净NPS推荐值第一名。 当下，享界四款车的产品矩阵则为品牌提供了持续覆盖细分需求的空间。同时，享界正在验证一个曾被反复质疑的判断：个性化或许本身可以成为一种高端化的路径。 享界进入每一个细分市场，都是在独立回答一个品类定义的问题。但这种策略无法借用已验证的产品模板，每一个品类都要独自承担市场教育的成本。但当四款车在同一个品牌下集结时，用户对享界的认知开始形成一种稳定的预期。 这种预期一旦建立，就不再需要品牌靠降价来驱动用户选择。在高端新能源汽车的牌桌上，建立在产品哲学上的品牌认同，可能是比销量更持久的壁垒。 文章原文

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

最几天跟游戏行业的朋友聊天，我总免不了问一句：“欸，你们有人去科隆了吗？”得到的也大多是肯定的答复。 当地时间8月26日至30日，gamescom 2026将在德国科隆举行。在北美E3展会宣告停办后，gamescom科隆游戏展已成为全球规模最大、影响力最强的综合性游戏展会。 8月25日，开幕夜直播活动Opening Night Live将先行登场。各大游戏厂商公布新作、公布首发消息，自然是每届gamescom关注度最高、流量最大的环节。哪些作品值得期待，又会有哪些意外的惊喜？ 去年的ONL，主持人还是Geoff Keighley 但如果把视线从新游新作稍微移开，在全球游戏行业的背景下，今年的gamescom正在试图回答一个更大的问题：当游戏行业经历过裁员、开发成本上涨、AI冲击和市场增长放缓之后，游戏还能为未来提供什么？ gamescom 2026给出的答案，写在它的主题里——“Games spark excitement for the future”，即“游戏，激发对未来的热情”。 在昨天结束的gamescom dev开幕致辞上，行业公益社区ASGC的发起人、腾讯游戏商务总监Amir Satvat以“游戏行业的重置：现状与未来”为主题做了一场演讲，呼吁行业不要想着“重返过去”的繁荣幻象，而是要“回到未来”：重新定义成功，接受并建立一个规模更小、更克制，但能够支撑长久职业生涯和持续产出优秀作品的健康生态。 过去几年，游戏行业讨论最多的词往往是增长、商业化、长线运营、裁员、成本控制和AI。相比之下，“未来”是一个听起来有些过于乐观的词。但至少从今年的gamescom来看，这种乐观并非没有依据。 展会官方公布的规模已经达到历史新高：超过1600家展商来自67个国家，其中今年参展的中国游戏厂商达60家，参展规模又一次创下新高；本次官方展览总面积达到23.3万平方米，并且首次全部售罄。与此同时，独立游戏区域共有超过420家公司参与，比2025年增长16%，同样创下历史新高。 gamescom官方微博确认有“60家中国参展企业” 全球游戏市场未必已经全面回暖，但至少gamescom正在变得比过去更热闹——游戏行业依然愿意聚集在一起，展示新的产品、新的技术，也重新讨论游戏的下一种可能。 ■ 亮点：AI与独立游戏 AI仍然是绕不开的话题。今年gamescom及其开发者大会gamescom dev对AI的关注，也明显从前几年“要不要用AI”的争论，转向了更加具体的生产实践、游戏体验和风险控制：AI究竟应该进入游戏开发的哪些环节，又会怎样改变游戏开发本身？ 在今年的专题设置中，AI也被单独划分为“AI in Gameplay”和“AI in Production”两个专题，分别讨论AI如何进入游戏体验和生产流程。前者关注AI与游戏设计、玩法和交互之间的关系，后者则更多讨论AI在实际开发和生产环节中的应用。 gamescom dev对于AI议题的专门介绍 值得一提的是，在AI专题下，腾讯带来了两场非常具体的技术分享。其中一场的主题是《从脚本到 AI Agent：复杂游戏的可扩展自主测试》，讨论如何将传统依赖固定脚本的自动化测试升级为AI Agent驱动的自主测试。AI在这里已经不是一个概念展示，而是开始进入实际的游戏QA流程。 另一场则聚焦《洛克王国》的技术升级，分享了团队如何利用全局光照、神经渲染等技术，将这款经典IP从页游升级为64平方公里的开放世界跨平台游戏，并解决昼夜光照、内存、带宽和多平台性能等问题。 同样值得关注的是，AI并不只出现在开发者大会的技术讨论中。gamescom congress今年同样将AI驱动的游戏开发列入议程，并把AI与沉浸式技术、教育、网络安全等议题放在一起讨论。由此来看，AI在今年gamescom讨论中的位置，已经不再只是一个单纯的生产工具，而开始延伸到游戏产业、技术乃至社会文化层面。 2025年的gamescom现场 而这恰好与今年gamescom强调的另一个趋势“Small teams, big success”（小团队，大成功），产生了联系。 AI工具正在降低部分游戏开发环节的技术和生产门槛。对于规模较小的团队来说，这意味着过去需要更多人力和时间完成的部分工作，可能出现新的生产方式。但这种变化最终会如何影响游戏团队的规模、工作方式，以及游戏本身，目前仍然是一个开放的问题。 至少在今年的gamescom现场，小团队能够获得比过去更多的关注。 2026年的gamescom indie area共有超过420家公司参与，比去年增长16%，达到历史最高水平。官方同时公布了一组有意思的数据：德国目前约有1700万玩家会游玩独立游戏，82%的玩家认为，小团队能够利用现代技术创造出这样的作品“值得赞叹”。 这一点也体现在今年gamescom Award的提名中。例如，由西班牙独立团队Undercoders开发的“电车动作”游戏《电车炫客》，一举获得4项提名；由法国独立游戏工作室The Game Bakers开发的生存攀岩游戏《孤山独影》，此次也获得了Best Gameplay和Best Audio两项提名。同样获得提名的《血墨诡案》和《DEPERSON》也都是具有相当辨识度的独立作品。 《电车炫客》的背景是一个日式反乌托邦世界 甚至还有一些更加“小”的项目。《Memoria Wake》的开发者蛋泥就在小红书上分享了自己的经历：这款游戏最初只是他和几位大学同学出于兴趣做的业余项目，团队成员此前几乎没有游戏制作经验，甚至连Unity都是临时开始自学。但几年之后，他们带着自己的作品登上了gamescom的Indie Arena Booth。 今年gamescom甚至专门加强了线上indie展示。主办方与IGN合作制作“gamescom awesome indies show”，通过预告片首发、实机演示和开发者故事，为独立游戏提供更多国际曝光。gamescom负责人Tim Endres在接受Wccftech采访时也提到，越来越多独立游戏正在获得全球玩家的关注，展会希望通过线上活动建立一个与indie area相对应的数字平台。 但这并不意味着gamescom正在从“大厂的展厅”变成“独立游戏的舞台”。 Tim Endres在采访中也指出，虽然今年indie area增长了16%，但真正占据大量展览空间的，依然是拥有大型展台的游戏厂商。一边是依然需要巨大成本和空间的3A与商业游戏，另一边则是越来越多能够用更小团队探索新玩法的独立作品。而AI正在进入其中，成为影响两者生产方式的一个新变量。 这或许也是今年gamescom把“小团队，大成功”摆出来的意义：游戏行业并没有因为开发成本越来越高，就只能走向更大的团队和更高的投入。至少从今年的展会上看，另一种可能性正在变得越来越清晰——用更小的团队，借助新的技术，去做过去很难做出来的游戏。 gamescom团队，Tim Endres站在前排右三 ■ 聚焦：中国游戏公司 昨天，Amir Satvat在gamescom dev开幕致辞上，提到了一些关于中国游戏市场的数据。2025年全球PC游戏收入增长中，大约42%来自中国。中国已经成为全球PC增长的中心驱动力之一。而根据ASGC模型估算，与北美游戏劳动力市场在2022至2026年间预计收缩约15%形成对比，中国市场的增量显得尤为突出。 到了gamescom现场，中国游戏公司的存在感也达到了一个新的高度。gamescom官方微博今天确认有“60家中国参展企业”，这一数量也创下了历史新高，相较于2023年的30家参展数量增加了整整一倍。从头部厂商，到中型团到，再到大量进入B2B展区的中小开发者，中国游戏公司正在以越来越完整的方式出现在gamescom的展馆里。 头部厂商的存在尤其明显。 腾讯在今年gamescom的参展投入规模空前。腾讯的to B展位面积达到700平米，加上其发行品牌Level Infinite的展区，整体规模超过1000平米。参展游戏阵容涵盖多个品类，包括《粒粒的小人国》《王者万象棋》、多款王者荣耀IP作品以及《洛克王国：世界》等等。其海外发行品牌Level Infinite带来了四款可试玩的重磅作品，包括《控制：共振》《沙丘：觉醒》以及《猎杀：对决1896》等等。由腾讯投资的国产单机游戏《湮灭之潮》也确认参展，并提供海外首次线下试玩。 除了主展位，腾讯还联合全球多家工作室，在科隆同期举办了名为“Worlds of Play: The Game Art Exhibition”的共创游戏艺术展。该展览于8月27日至29日举办，聚焦40余款经典游戏及系列作品，包括“刺客信条”系列、《博得之门3》《黑神话：悟空》《影之刃零》等等中国与海外知名作品。腾讯正在尝试把游戏IP、艺术展览和全球游戏文化放到同一个场景里。 “Worlds of Play: The Game Art Exhibition”共创游戏艺术展 网易这次则带来了《漫威争锋》《燕云十六声》《七日世界》《遗忘之海》等产品，其中《漫威争锋》和《燕云十六声》位于7.1馆的消费者展区，《遗忘之海》拥有独立展位。与此同时，网易还在4.2馆设置了自己的商务展区，并为《七日世界》和“一款新作”提供现场试玩和媒体采访——结合此前《无限大》公布的参加ONL信息，这里的新作大概率会是这款备受期待的二次元开放世界都市游戏。 除此之外，米哈游、鹰角、叠纸等等也都分别带来了各自的旗舰作品以及新作。值得一提的是，在昨天宣布的gamescom游戏奖的提名中，入围的五款“最佳移动游戏”：《明日方舟：终末地》《原神》《白银之城》《燕云十六声》和《绝区零》均来自中国游戏公司。 中国游戏包揽了“最佳移动游戏”提名 而今年另一个很直观的变化是，中国游戏厂商的存在已经开始从“产品展示”向“行业交流”延伸。在gamescom dev，我们能看到越来越多的中国开发者的身影。 除了在前面提到的腾讯带来的分享外，今年gamescom dev首次设置专门的Level Design Track，讨论世界构建、玩家引导、环境叙事等关卡设计问题，其中有一场是来自叠纸的《无限暖暖》开发团队：他们讨论了团队如何面对移动端性能限制、开放世界持续更新和紧张的GaaS开发周期，通过技术迭代和自动化系统提升音频生产效率。 或许几年前，中国游戏公司参加gamescom，更多被理解为“把国产游戏带到欧洲”。而现在，越来越多中国公司已经把这里当成全球发行、品牌建设和行业合作的一部分。中国游戏公司已经很难再被视作gamescom上的“海外来客”，而正在成为这个全球游戏市场的重要组成部分。 ■ 未来 距离gamescom 2026真正向玩家敞开大门，只剩下几个小时。 当然，我们非常期待明天凌晨的Opening Night Live：期待那些尚未公布的新作，期待久违的老IP，也期待某个完全没有预料到的惊喜。对于一场游戏展来说，这些永远是最直接、也最令人兴奋的部分。 不过对于游戏行业而言，也许比“今年有什么新游戏”更值得观察的是：在经历这一轮行业变化之后，游戏会长成什么新的样子。 我们不指望一场游戏展能给出多么明确的答案，但至少在接下来的这几天里，我们会尝试去寻找其中的蛛丝马迹。

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Earlier today, Apple unveiled the new generation of Mac mini and Mac Studio, featuring the latest entries in the Apple silicon family of chips: the M6, available in the Mac mini, and the M5 Ultra, exclusive to the Mac Studio. You can read more details about the announcement and related specs in John’s overview . As someone who’s been working with an M3 Ultra Mac Studio (on loan from Apple) with 512 GB of RAM for the past year (plus two separate M4 Mac minis) with a particular focus on local AI models and performance gains enabled by Apple’s MLX framework, I obviously am very interested in these new machines. To give you some context: my current workspace for my upcoming iOS and iPadOS 27 review, which lives in Notion, is entirely managed by a series of local agents running the latest DeepSeek-V4-Flash via MLX on the Mac Studio, which continuously scan the project for new notes, sources, and research material that is automatically categorized and linked in the chapters that I’m writing. So, yes, I’m sure I’ll have more thoughts on the new Mac minis and Mac Studios soon. In the meantime, I thought it’d be fun to break down the local AI-related details from Apple. First, the M6. While it may not be the powerhouse that the M5 Ultra is, it’s an interesting new chip because of its new 2 nm process and dual Neural Engine system: M6 is built using cutting-edge 2 nm process technology, packing greater transistor density into a smaller die for a major leap in performance and power efficiency. M6 also introduces a Dual 16-core Neural Engine, providing up to 2x the peak compute over previous generations to make on-device AI workflows run even faster. System frameworks can automatically utilize both engines simultaneously, enabling applications to see faster model execution. But that’s not all: the new 12-core GPU (which has two more cores than the M4’s old GPU) now features Neural Accelerators in each core for the first time on the Mac mini. The result, according to Apple, is “4x faster AI performance and 2x faster graphics than Mac mini with M4”. If these numbers hold up and scale linearly, a local Mixture-of-Experts model such as Qwen 3.5-35B-A3B , which would run at ~17 tokens/sec on average on a base M4 Mac mini with 16 GB of RAM, could realistically generate output at over 60 tokens/second with the base model M6 Mac mini. Based on what we’ve seen so far, the one downside of the M6 Mac mini is that it does not have Thunderbolt 5 ports; those are exclusive to the M5 Pro model, also announced today. As Apple also notes, Thunderbolt 5 allows multiple M5 Pro Mac minis to run local AI models as a single cluster (for example, using something like exo ). Alas, that won’t be possible with the M6 Mac mini and its Thunderbolt 4 bandwidth bottleneck. Nonetheless, I’d be intrigued to compare the performance of the widely regarded Qwen3.8-27B and DeepSeek-V4-Flash models on the new M6 Mac mini. Moving on to the flagship M5 Ultra Mac Studio, let’s start here: With M5 Ultra, Mac Studio achieves up to 4.3x the peak AI compute performance of M3 Ultra and a staggering 9.8x more than M1 Ultra. Combined with up to 512GB of unified memory and 1.2TB/s of memory bandwidth, 50 percent higher than before, Mac Studio lets users run massive models entirely on device with complete privacy — without counting tokens or worrying about rising cloud costs. As seen on the Mac Studio’s product page, the 512 GB RAM version of the M5 Ultra Mac Studio will only be available in “late October”. Unlike the M6, the M5 Ultra supports Thunderbolt 5, which will allow customers to run multiple Mac Studios in a cluster with shared memory to distribute inference across machines. And, according to Apple, a cluster of four Mac Studios yields three times the performance of a single Mac Studio. Speaking from personal experience, I know that my M3 Ultra Mac Studio using oMLX can run DeepSeek-V4-Flash locally with generation averaging 35 tokens/second. Assuming a linear 4x increase, that would put the same model at over 120 tokens/second on an M5 Ultra Mac Studio. To put things in perspective, that kind of performance would be faster than any AI chatbot website, it’d be faster than many providers who offer a “fast” mode for their models, and it’d only be second to either dedicated NVIDIA PC clusters at home or specialized inference providers such as Cerebras or Groq …which are running in full-blown data centers. Sure, you would need a computer that is likely going to cost more than $20,000 to make it happen, but it’d still be possible on a single machine that is small, quiet, and that – in theory – any consumer can buy off the shelf. It’s also interesting to consider the implications of M5 Ultra for running much larger models that would have typically required aggressive quantization to even stay in memory and produce any useful output. For example, Z.ai’s GLM-5.2 (a 744B MoE model) could run at up to 17 tokens/second on an M3 Ultra Mac Studio with 512 GB of RAM; the same person was even able to get the massive Kimi K3 model to run on a single machine…with decoding performance up to 3 tokens/second. You see where this is going: by continuing to push the envelope of what is possible with M-series chips, Apple is gradually making it possible to run frontier open-weight models on local hardware that takes up as much space as four Mac minis stacked together. Obviously, for now, this is all theoretical: we’re going to need actual benchmarks to objectively measure the improvements and performance gains of the new chips. But if Apple’s claims hold up – and I have no reason to believe they won’t – I believe both the M6 and M5 Ultra will meaningfully shake up the landscape of local AI models on macOS, reinforcing , once again, that Apple continues making the best computers for AI in the industry.

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Macworld As predicted on this site earlier today, Apple has announced the launch of a new Mac mini model with M5 and M6 chip options. It doesn’t feature any design changes or external upgrades, but it does offer a two-generation bump in processors compared to the previous model. Slightly confusingly, the version of the Mac mini with a chip from the M5 generation is the more powerful of the two, because it’s the M5 Pro . This spec starts at $1,699 for a 15-core CPU and 16-core GPU, but customers have the option to bump that to 18 and 20 cores, respectively, for an additional $200. There are also configuration options to increase RAM from the starting 24GB or storage from the baseline 512GB, and to upgrade to 10 Gigabit Ethernet. The RAM options remain the same, but the 64GB highest-end option now costs $1,000. More appealing for the Mac mini’s traditional budget market will be the M6 model, which starts at a far more manageable $899. That, however, is still a $100 increase over the M4 model from 2024 that it replaces. Worse, the M4 model started at just $599 until June, when Apple responded to the ongoing memory crunch by imposing brutal price hikes . In other words, the cheapest entry point to Apple’s desktop Mac ecosystem has jumped by $300, or 50 percent, in the space of two months. The $899 version isn’t exactly a state-of-the-art build, either. The entry-level configuration comes with 16GB of RAM and just 256GB of storage. We should note that it is possible to get this spec for $799, but only if you’re an education buyer . Of note, the Mac mini is not available for the Apple Upgrade program either. Apple The price hike is likely to weaken the Mac mini’s appeal among budget buyers, but (aside from the necessity to accommodate increasing component costs) Apple may not care. In his report predicting that a new mini was on the way, Bloomberg’s Mark Gurman noted that the machine had become increasingly popular for running AI applications locally, and that Apple was struggling to meet demand. The company may not mind if fewer customers are prepared to pay the asking price, since there are plenty of AI fans ready to take their place. Those AI fans will likely be tempted by a new model with a new chip. Apple boasts that the M6 delivers “gains across every dimension of performance” and “the ideal balance of performance, power efficiency, and on-device AI to effortlessly fly through daily tasks, coding, and creative projects.” In numerical terms, the company claims the M6 version of the Mac will deliver up to 4x faster AI performance, 2x faster storage and graphics, and 40% faster CPU performance than the M4 mini. The M5 Pro model, meanwhile, is rated by its maker as up to 8.5x faster at LLM prompt processing than its M2 Pro predecessor, and up to 4x faster than the M4 Pro model. Naturally we will put those claims to the test in our labs. Another surprise is how early Apple has announced the new Mac mini. Pre-orders are open now, but it won’t go on sale until 22 September, almost a month from now.

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

最近重构一个项目，sol5.6 老降智商，防御性编码太多，设计出来的东西也要跟他掰扯很久才能如意 第一次 想想开个 Claude 试试，AppStore 付款，爽蹬一个下午，感觉还挺耐用，一个半小时差不多跑完了五小时限额的 80%。 然后喜提 ON HOLD 。很快啊，写了解封邮件。calude 你不讲武德啊，我正儿八经搞项目呢，你给我封了是弄啥嘞。 然后丝滑小连招，appleStore 同步申请退款。 过了一天 apple 退款到账，第二天 calude 回复 remain ON HOLD 。不给我解封是吧，你等着 第二次 然后又注册了一个号 这次做足准备，查了 calude 封号原因，ip 是很大一个因素，重金上车了美区 DMIT EB VPS 。号称家宽优化 那顶的确实比第一次久一点，足足用了三天！不过我分析可能是头两天没怎么用，今天重度使用了一下，直接又 ON HOLD 了， 我淦霖娘啊 CLAUDE ！ 让 gpt 帮我写了解封申请，不过已经不抱什么希望了，就这样吧 bro 。 最后 折腾了一圈，用回了 GPT 、感觉也不是那么差嘛。 claude 不让用，咱也没必要上赶着给他送钱 自此 openAI 又眉清目秀了起来 一切又回到了原来的起点... (多写点 promote 就好啦

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Engineering students rarely work in one place. They bounce between lectures, labs, and libraries. That means your laptop has to be lightweight, long-lasting, and ready for heavy workloads. The options below balance portability with the power needed for CAD and simulations. Why you should trust PCWorld for laptop reviews and buying advice: It’s in our name! PCWorld prides itself on laptop experience and expertise. We’ve been covering PCs since 1983, and we now review more than 70 laptops every year. All of the picks below have been personally tested and vetted by our experts, who’ve applied not only performance benchmarks but rigorous usability standards. We’re also committed to reviewing PC laptops at every price point to help you find a machine that matches your budget. Asus Zenbook 14 OLED – Best overall Pros Robust build quality Attractive OLED touchscreen Good CPU and integrated GPU performance Outstanding battery life Cons Design is dull Keyboard isn’t memorable Mediocre connectivity Price When Reviewed: $849.99 Best Prices Today: Retailer Price 10% off original price$759 View Deal Asus $999.99 View Deal $1,289 View Deal Price comparison from over 24,000 stores worldwide Who should buy the Asus Zenbook 14 OLED? The Asus Zenbook 14 OLED tops our list because of the brilliant combination of power, portability, and battery life. The AMD Ryzen 7 chip is designed to run heavier programs like CAD without slowing down, even if you’ve got a bunch of tabs open on the side. The 14-inch 1920×1200 OLED touchscreen also looks fantastic and will make your diagrams and other projects really pop. The one thing that really sticks out is the battery, though. Seriously, it went for over 16 hours in our test, which kind of blew my mind. You can definitely survive a whole day of classes, especially with conservative use and the brightness turned down. The audio is also quite good, offering you good volume and clarity during remote lectures. Asus Zenbook 14 OLED: Further considerations The build quality’s pretty good, especially considering it’s just $699 as of this writing. It’s also thin (0.59 inches) and light (2.82 pounds), which is great if you’re running from class to class. While the keyboard isn’t the best we’ve ever used because of the undersized keys, it’s totally fine for day-to-day use. Read our full Asus Zenbook 14 OLED review Dell 14S – Best battery life Pros Class-leading battery life Slick build Colorful and accurate display Good port selection Cons Dismal 3D performance Display isn’t very bright So-so performance for the money Price When Reviewed: $1,469 Best Prices Today: Retailer Price 13% off original price$1269.99 View Deal Price comparison from over 24,000 stores worldwide Who should buy the Dell 14S? The Dell 14S is a great fit for engineering students who want a laptop that can easily last through a long day of classes. It actually delivered some of the longest runtimes we’ve seen (nearly 36 hours in our video playback test!), thanks in part to its efficient Intel Core Ultra 7 355 processor and 69 watt-hour battery. It’s also just a solid day-to-day machine. The aluminum design feels nice and sturdy, the keyboard is comfortable to type on for long periods of time, and the laptop remains easy to carry at just 3.28 pounds. For students who mainly use office apps, the Dell 14S has more than enough power for that kind of thing. Dell 14S: Further considerations This isn’t the laptop to buy if you need raw performance. The Dell 14S relies on Intel integrated graphics rather than a dedicated GPU, so it’s not the best choice for running demanding CAD projects. Ultimately, the Dell 14S is a portable, reliable machine for students who spend most of their day away from a desk. It puts efficiency first. Read our full Dell 14S review Acer Nitro V 16 AI – Best for work and gaming on a budget Pros Large touchpad compared to alternatives 16-inch 1200p display at a low price Lots of physical connectivity Good battery life for a budget gaming laptop Cons Sub-par CPU performance Terrible audio quality from built-in speakers Wireless connectivity stuck on the Wi-Fi 6 standards Slightly more expensive laptops offer better value Price When Reviewed: $899 Best Prices Today: Retailer Price Check on Amazon Price comparison from over 24,000 stores worldwide Who should buy the Acer Nitro V 16 AI? If you’re an engineering student who likes to game after class, the Acer Nitro V 16 AI is worth considering. The RTX 5050 tucked inside delivers smooth 1080p gaming, hitting a respectable 72 FPS in Cyberpunk 2077 on Ultra settings. It also gives extra firepower for CAD modeling and rendering tasks. The 16-inch 1920×1200 IPS display is slightly taller than the typical 15.6-inch 1080p screens, so you’re getting more vertical workspace here–this is useful for coding, spreadsheets, and so on. At a price hovering between $800 and $900 it’s a budget-friendly option for students. Acer Nitro V 16 AI: Further considerations The AMD Ryzen 5 240 isn’t the fastest CPU around, and the speakers aren’t impressive, but despite these trade-offs, the Acer Nitro V 16 AI avoids common pitfalls most budget gaming laptops fall into. It lasted 12 hours on a single charge in testing, which is impressive for a gaming machine. The keyboard and touchpad are also spacious, making it comfortable to work on for long periods. Read our full Acer Nitro V 16 AI review Microsoft Surface Pro – Best ultraportable Pros Compact and well-built Good office performance All Copilot Plus features available Good battery life Cons Narrow screen to work on No HDR support No charger included Price When Reviewed: $799.99 Best Prices Today: Retailer Price $949.99 View Deal $949.99 View Deal $949.99 View Deal $969.99 View Deal $1049.99 View Deal $1049.99 View Deal Price comparison from over 24,000 stores worldwide View more prices Who should buy the Microsoft Surface Pro? If you’re an engineering student with a jam-packed schedule, often needing to run from class to class, the Surface Pro is so lightweight (just a little over a pound) you’ll hardly notice its presence. Inside this laptop you’ll find a Snapdragon X Plus chip, so you can expect it to deliver a quiet, fanless experience (perfect for working in libraries and lecture halls), and it should handle basic coding and homework just fine. Note: Snapdragon X Plus / Arm models are fine for lighter coding and notes, but if your coursework is heavy on SolidWorks, Autodesk apps, or older MATLAB workflows an x86 Windows laptop is a safer bet for performance and app compatibility first. The 12-inch 2196×1464 screen offers sharp colors and touchscreen capabilities, allowing you to draw and take notes with the (optical) Slim Pen. Microsoft Surface Pro: Further considerations Though we really vibe with the high-resolution display, it’s a bit small if you’re the kind of person that has many windows open at once. Battery life is all over the place, ranging from two to three hours under heavy loads to 19 hours with light usage. You’re also limited to just two USB-C ports, so you’ll probably want to pick up a dock or hub. Read our full Microsoft Surface Pro 2025 review Other notable laptops we’ve tested Geekom GeekBook M16 : The GeekBook M16 is a solid budget productivity laptop with outstanding battery life and build quality, but its weak display and limited graphics performance hold it back from being a true all-rounder. Dell XPS 13 : Dell’s XPS 13 is a compelling budget laptop, even if it falls just slightly behind the MacBook Neo. Asus Zenbook A16 : A solid productivity laptop, with top-notch office performance, an outstanding multimedia experience, and excellent battery life MSI Prestige 16 Flip AI+ : The MSI Prestige 16 Flip AI+ is a great laptop with a quality build, zippy speeds in everyday operation, and a wonderful display for work and play. How we test laptops The PCWorld team puts each and every Windows laptop through a series of benchmarks that test GPU and CPU performance, battery life, and so on. The idea is to push the laptop to its limits and then compare it against others we’ve tested. Chromebooks, on the other hand, go through a series of web-based tests. For a much deeper look at our review methodology, check out how PCWorld tests laptops . Who curated this article? Hi, I’m Ashley Biancuzzo , and I oversee all laptop and Chromebook coverage at PCWorld. While you’ll see me review Chromebooks on occasion, I’m also really into the broader world of consumer tech. I spend a lot of time writing and thinking about where laptops are headed–from AI and sustainable designs to long-term trends. When I’m not deep in the world of tech, you’ll probably find me gaming, getting lost in a good book, or chilling with my rescue greyhound, Allen. How to choose the best laptop for engineering students The following are important questions to ask before making a purchase. How much power do you need? You’ll need a good amount of power. Engineering classes usually have you running some pretty heavy stuff like CAD, MATLAB, SolidWorks, and simulations. That stuff can slow your laptop down if it’s not up to the task, especially if you’ve got a bunch of apps open. A Core i5 or Ryzen 5 will get you through most of it, but if you want things to run smoother, a Core i7 or Ryzen 7 is where it’s at (so long as your wallet agrees). Is RAM important? For sure! You’ll be using code editors, compilers, simulation tools, sometimes all at once. RAM keeps your laptop from freezing up when you’ve got a ton going on. And with RAM prices being a bit volatile nowadays, it’s worth future-proofing if you can. 8GB might work if you’re just starting out, but 16GB is way better, especially later on when your classes get more intense. How much storage should you get? You’ll collect a ton of files–project folders, software installs, maybe some video editing or game dev work depending on your focus. We’d say go for at least 512GB SSD so you’re not constantly clearing space. If you can get 1TB, do it. It’s faster and gives you more breathing room for whatever your classes (or side hobbies) throw at you. FAQ 1. What is the best laptop for engineering students? If you’re an engineering student who needs something light, fast, and reliable for coding or writing papers (plus a great screen and battery), the Asus Zenbook 14 OLED is a dang fine pick. 2. What is the best budget laptop for engineering students? If you’re on a tight budget and just need a straightforward for lighter coding programs, the Acer Aspire Go 15 gets the job done for just under $500. 3. Is battery life important for engineering students? Heck yeah! Long battery life means you can power through those long nights at the school library. 4. What’s the ideal screen size for an engineering laptop? We’d say the best screen size is anywhere from 13- to 15-inches. It can’t be too cumbersome otherwise it’ll get difficult to carry around. Related content PCWorld laptop reviews What’s new when shopping for a laptop in 2026? 8 things to keep in mind When PC vendors tell you what’s going on with tariffs, pay attention How to try a new laptop at home before you commit to buying it How to properly wipe your laptop data before getting rid of it 4 eco-friendly ways to get rid of a laptop Battery life on gaming laptops doesn’t have to suck anymore Best gaming laptops under $1,000: Expert picks that won’t break the bank

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Okunbowa For millions of Nigerians, reforms of the past three years have been experienced less as policy than as pressure. They have been felt in higher petrol prices, rising food costs, more expensive credit and a sharply adjusted naira. For businesses, the transition brought higher operating costs and pricing decisions. For households, it tested stretched incomes. Yet the Federal Government’s Reform Scorecard suggests that beneath these pressures, parts of the economy are beginning to stabilise. At the center of the adjustment was the removal of the petrol subsidy and the dismantling of a fragmented foreign-exchange system that encouraged arbitrage while placing growing pressure on public finances. Between June 2023 and December 2025, subsidy savings generated ₦15.8 trillion for the Federation. Of this, ₦5.4 trillion accrued to the Federal Government, while states and local governments received a combined ₦10.4 trillion. That distinction matters because the reforms have altered the fiscal position of subnational governments. In 2023, 27 states could not reliably meet salary obligations. By 2026, the scorecard records none in that position. For civil servants, pensioners and businesses dependent on government activity, stronger cash flow reaches far beyond accounting tables. At the federal level, the reforms created additional room for expenditure during considerable economic strain. The government recorded ₦20.4 trillion in incremental resources, comprising ₦5.4 trillion from its share of subsidy savings, ₦3.1 trillion in other incremental revenue and ₦11.9 trillion in borrowing. How that money was deployed is equally important. Of ₦30.64 trillion in incremental expenditure, ₦9.39 trillion went to wage adjustments, minimum-wage increases and allowances, while ₦6.47 trillion was directed to strategic infrastructure. The increased naira cost of servicing external debt absorbed ₦9.37 trillion. Federal wage adjustments alone exceeded the Federal Government’s subsidy savings, while the national minimum wage rose from ₦30,000 to ₦70,000. Other indicators point to an improving fiscal position. Debt service, equivalent to 100 per cent of revenue in 2022, is projected at 50 per cent in 2026. Ways and Means financing, with a legacy stock of ₦30 trillion, has been curtailed, while Nigeria’s tax-to-GDP ratio has risen from below 10 per cent to around 12.5 per cent. External buffers have strengthened considerably. Gross foreign reserves rose from approximately $35 billion in May 2023 to $52.5 billion by July 2026, while net reserves increased from about $3 billion to $34.8 billion. The gap between the official and parallel foreign-exchange markets, previously above 60 per cent, has narrowed to less than 5 per cent. Capital flows are responding. Total capital importation increased from $1.13 billion in the first quarter of 2023 to $10.37 billion in the first quarter of 2026. Foreign direct investment rose from $895 million in 2022 to $4.01 billion in 2025, while stock-market capitalisation expanded from about ₦31 trillion to roughly ₦150 trillion by June 2026. The productive economy is also showing signs of improvement. Real GDP growth increased from 2.31 per cent in the first quarter of 2023 to 3.89 per cent in the first quarter of 2026. Non-oil growth is projected at about 4 per cent, while manufacturing expanded by 3.29 per cent. Still, the reform story remains incomplete. Petrol sells for roughly ₦1,100 to ₦1,400 per litre. Monetary policy remains tight, with the policy rate at 26.5 per cent. Poverty remains high and household welfare is classified as a work in progress. Food inflation has eased from 24.82 per cent to 17.52 per cent, but that does not mean food has become cheap; it means prices are rising more slowly. That distinction defines the next phase of Nigeria’s reform story. The administration points to stronger reserves, improving growth, healthier fiscal ratios, recovering capital flows and a more functional foreign-exchange market. These are foundations. But macroeconomic repair becomes durable only when it changes household circumstances through purchasing power, jobs and lower living costs. The numbers suggest that stabilisation is taking hold. The task now is ensuring Nigerians can feel it in daily life. Osaro Okunbowa: Beyond the pain – What Nigeria’s reform numbers are beginning to show

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

本文自南都·湾财社 采写 | 南都·湾财社记者 詹丹晴 河北张家口康保县“甲醛白菜”事件引发关注。8月22日，博主“渔猎齐哥”发布的视频显示，在河北张家口市康保县屯垦镇，工作人员将大白菜装载进货车前，会先蘸取甲醛溶液用于保鲜，有货车司机称，这批白菜将运往江苏宿迁。 随后，康保县政府发布情况通报称，经初步核查，视频反映情况属实，系收购商为蔬菜运输保鲜，在装运过程中的违法行为。 8月23日晚，国家市场监督管理总局发布消息称，针对该事件，国务院食安办、农业农村部、市场监管总局高度重视，第一时间指导属地调查核实，依法从严从快处置，对涉事白菜流向进行追查，严防问题产品流入市场。同时举一反三，组织各地立即开展白菜等易腐蔬菜专项抽检和市场排查，一旦发现违法违规行为，将严惩重处、决不姑息，切实维护人民群众饮食安全。 “甲醛白菜”引发关注，多地消费者担心会买到“甲醛白菜”。 8月25日，康保县市场监督管理局工作人员告诉南都湾财社记者，目前该局已经在运输路上设置了检测卡口，对蔬菜进行快检，检测合格的才会放行，检测不合格的则直接扣押。记者向其了解涉事的“甲醛白菜”是否仅流向安徽和江苏，该工作人员表示，目前没有进一步的官方消息，建议等官方最新的通报。 8月24日，广州江南果菜批发市场相关负责人告诉南都湾财社记者，涉事白菜并未流向江南果菜批发市场。江南市场严格把好进货关口，要求各档口严禁采购、销售涉事问题来源大白菜；所有白菜进场销售，须将购货凭证、运输单据、产地承诺达标合格证上传市场溯源系统，实现来源可查。并且对全场白菜品类开展高频专项抽检。 朴朴超市相关负责人向南都湾财社记者指出，朴朴平台未上架销售来自此次舆情涉事产地的白菜商品。平台通过严选规模大、资质齐全的合规供应商，驻点采购源头品控把关；蔬菜采收后物理预冷、全程冷链保鲜；除常规农残快检外，产地与仓库增设甲醛筛查，上架后不定期抽检，多重举措守护菜品安全。 华润万家相关负责人告诉南都湾财社记者，“经查，我司未经营产自康保县的大白菜。我司持续严格落实索证索票制度，确保销售的商品来源清晰、证照齐全、质量达标，从源头保障商品质量安全。”永辉超市相关负责人亦向南都湾财社记者指出，经确认，永辉超市全国门店并未从事发地采购大白菜。 南都湾财社记者亦就此事向叮咚买菜、盒马、小象超市、大润发等了解情况，截至发稿，尚未收到回复。记者在叮咚买菜平台上看到，其销售的大白菜主要来自甘肃省连搭乡、安徽蒙城县、山东红河镇；盒马App显示，广州市场销售的主要是娃娃菜。 END 文章原文

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

撰文 | Skin 下图的衣服是一家名为noRecognition的时尚服装，你觉得好看吗？眼下，这类花纹的衣服正在变得越来越流行。 许多人开始购买这类衣服，许多品牌开始推出这样的衣服。你可能觉得是“时尚”难以让人理解，不过，它们之所以能流行起来，其实是出于安全性的考量。 人们为这类衣服起了名字，叫做“对抗性服装”：一些让人眼花缭乱的彩色图案，是为了“对抗”AI识别算法。 计算机视觉算法正越来越多地出现在我们生活中，人们被拍下，识别。技术的进步虽然帮助了许多人，但算法也让人们陷入了隐私泄露的恐慌。近期的一个案例就提到，有人通过第三方非法网站购买了人脸照片，将其转化为“AI动态视频”，用此就能登录别人的互联网账号，骗取数万元财产。 另外，2024年在澳大利亚，一些商铺使用的面部识别系统遭到了黑客攻击，大量顾客的信息被盗。这些信息也很有可能被利用。就在一天以前，以为网络安全员发现，一个号称“私密安全”的软件，在网上有一个数据库，泄露了超过900万张图片。这意味着，骗子可以随时获取大量的面部信息。 隐私泄露风险和深度伪造技术，人们对自己“脸被偷走”的担忧与日俱增，而这类“对抗性服装”，或许成为了一种比较实用的解法。 担忧自己的脸被偷走，已经成为了一种与日俱增的焦虑。｜petapixel noRecognition的发起者是一名研究人员，他开发了一种对抗算法，能生成一些衣服上奇怪花纹，这些花纹会干扰视觉识别模型，让AI“什么都看不到”。这是因为，视觉模型处理图像的方式和我们并不同，它需要从像素中提取边缘、纹理、空间结构等特征，再依靠这些特征判断画面的具体内容。 一些特定的几何结构会影响模型对空间特征的提取，一些纹理会影响卷积网络早期的特征处理，还有些图案会干扰模型对人脸的识别。就算它只是一些花纹，这些干扰图案也会让AI难以识别眼前的图案。 noRecognition的研究测试表明，穿上“丑花纹衣服”，和穿上“普通黑灰色衣服”比起来，“丑花纹衣服”确实会让AI视觉模型认不出人脸。这些图案会让一些视觉模型出现漏检、错误识别的结果。 比起穿黑色衣服，如果穿着丑衣服能影响AI视觉模型的识别效果，就算是“攻击成功”了。｜noRecognition 在测试中，研究者用这种对抗算法生成的“丑图案”，测试了11个不同的视觉识别模型，结果发现，生成的图案可以让这些模型识别失败。不过值得注意的是，这些测试都是模拟的，并没有人真的穿着这些衣服。而noRecognition目前也在众筹，参与者日后将会获得自己独一无二的丑图案衣服。 这并不是第一个做对抗式服装的团队，目前市面上已经涌现了许许多多做“丑衣服”的公司。名为Cap_able的公司已经做出了一大批五颜六色的花式服装，宣称能让AI识别的难度加大。Urban Privacy推出了一批防拍照、防识别、防扫码的服饰…… Urban Privacy的各类“防识别”，“防拍照”衣物。 当然，有些服饰也许形式大于内容，但它们越来越流行，也并非不是好事。因为它持续提醒着人们，在数字时代，我们的形象已成为非常重要的财产，人们需要注重形象隐私，用各种新的方式来保证自己的安全。 名为Cap_able的公司做出了各种宣称能抵抗算法识别的服装，旨在提醒公众隐私保护的重要性。 参考资料 [1]https://sandbox.norecognition.org/research [2]https://blackhat.com/us-26/briefings/schedule/index.html#could-a-pattern-on-your-clothing-fool-facial-recognition-53532 [3]https://www.theguardian.com/fashion/2026/jul/17/adversarial-clothing-are-garments-designed-to-confuse-facial-recognition-systems-about-to-go-mainstream [4]https://news.cctv.com/2026/08/16/ARTIagm353GH4xMQx1ABnxwu260816.shtml [5]https://urban-privacy.com/collections/privacy-shop 本文转载自公众号“把科学带回家”（id：steamforkids） 凡本公众号转载、引用的文章 、图片、音频、视频文件等资料的版权归版权所有人所有，因此产生相关后果，由版权所有人、原始发布者和内容提供者承担，如有侵权请联系删除。 -电商广告- 《环球科学》2026年8月新刊销售中 戳图片或阅读原文 立即购买 点击 【在看】 ，及时接收我们的内容更新 文章原文 平台地址： http://www.jintiankansha.me/t/bfj2u79QHR

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

A Shakespearean saga is playing out between the White House, Treasury Department, the Federal Reserve and Wall Street—and Scott Bessent, to paraphrase Shakespeare, is being hoist on his own hedge-fund petard. As Hamlet told his mother Gertrude in Act 3, Scene 4, having just stabbed an eavesdropping Polonius, “’tis the sport to have the engineer/ Hoist with his own petard.” Now Bessent’s former mentor, Stanley Druckenmiller, is the one pulling the trigger—using the same playbook they wrote together over 30 years ago. In the early 1990s, hedge funds were evolving, and Bessent and Druckenmiller were there at the inception. Their boss, George Soros, pioneered a “ global macro ” approach that discovered sovereign balance sheets could be read the same way a company’s could: an investing opportunity for the gap between what a government claimed it could sustain and what the market would allow. The defining proof came in 1992, when Britain was maintaining the pound inside Europe’s exchange-rate mechanism at a level that German interest rates had made untenable. Soros Fund Management built a short position of roughly $10 billion against sterling; Druckenmiller ran the trade and a young Scott Bessent was part of the team. When the pound broke on September 16 , the fund made roughly $1 billion in a single day. Now Druckenmiller is invoking the same logic against Bessent, who has crossed from the trading desk to the Treasury Department. He used the Wall Street Journal opinion page to call out his former protege. But, perhaps unprecedentedly, he did so with an AI-assisted essay. Jeff Stein, the Pulitzer-winning former chief economics correspondent for the Washington Post, wrote on X that he contacted Druckenmiller, who responded “of course” he used AI to write the essay: “There’s a reason I moved from an English major to being an economics major. I’m not embarrassed by it.” Druckenmiller could not be immediately reached for comment by Fortune . The Treasury Department did not respond to a request for comment. In the Journal , Druckenmiller criticized Treasury’s decision to double long-dated bond buybacks from $2 billion to at least $4 billion per operation—operations targeting securities with maturities of 10 to 30 years, announced after the 30-year Treasury yield had reached a 19-year high. “The market’s verdict was swift and correct,” Druckenmiller wrote. “This wasn’t liquidity management, it was price management.” Jon Hilsenrath, who spent two decades covering the Federal Reserve and Treasury for the Journal , read Druckenmiller’s decision to publish as significant in itself. “The fact that he went to the Journal with it suggests to me that he didn’t think his message was getting through,” Hilsenrath told Fortune . He also noted that after serving as Bessent’s mentor at the Soros Fund, Druckenmiller later got closer to Federal Reserve Chair Kevin Warsh. The situation has a Shakespearean shape—the master watching two proteges navigate a principal whose economic instincts run contrary to what he taught them. Put that way to Hilsenrath, he didn’t resist the framing. “Druckenmiller’s two most prominent students are now running economic policy,” he said, one at Treasury, one at the Fed, “and they are doing so for a president who has a completely different worldview.” Druckenmiller, Hilsenrath noted, didn’t mention Trump by name in his op-ed. The omission is deliberate: the piece puts Druckenmiller at odds with Bessent without putting him openly at odds with the president. The alignment between the two proteges may be less complete than it appears. Warsh has articulated a market-purist position: let yields speak, don’t intervene. Bessent’s stated rationale for the buyback expansion is nearly its opposite—that Treasury has asymmetric information about market functioning and should act on it. “Those are two diametrically opposed views of the world,” Hilsenrath said. It matters, he added, because budget deficits are “clearly out of line with what the fundamentals say they should be,” and every American is paying the price. What the long bond says Druckenmiller’s argument is not that Treasury can never buy back securities. The modern buyback program was introduced in 2024 as a tool for liquidity and cash management. Buying older, less actively traded “off-the-run” bonds can improve market functioning without attempting to dictate the level of yields. His argument is about timing and presentation. Treasury enlarged the program after the 30-year yield hit a two-decade high, outside the usual quarterly-refunding rhythm, and Bessent subsequently suggested it could grow further. To Druckenmiller, that is the difference between debt management and price management. He saw no failed auctions, dealer-balance-sheet seizure or forced unwind of the kind that accompanied Treasury-market turmoil in March 2020 or the U.K. gilt crisis of 2022. He also contended that buying longer-dated debt while funding purchases with bills shifts duration risk out of private hands—a limited form of easing undertaken by Treasury rather than the Federal Reserve, and a problematic one when inflation remains above the Fed’s target. Treasury can offer a different account: properly designed buybacks are a routine, bounded technique for improving liquidity and managing cash, not a formal cap on yields or a covert monetary-policy tool. But the distinction is perishable. If investors read the Aug. 19 decision as Treasury flinching at an unwelcome price—rather than responding to genuine market dysfunction—it invites further tests of official resolve. Asked to calibrate the danger, Hilsenrath was measured. “A 5% Treasury yield is not a clear and present danger to the economy,” he said. “But it is a problem, which is why you have to pay attention to these market signals now.” Hilsenrath added that his own view is that the bond market has been “complacent” for a long time, and maybe, per Druckenmiller’s point, “the bond market might just now be waking up.” The industry that changed The warning lands differently because the Treasury market Bessent is managing is not the one that financed America’s deficits when Druckenmiller and Soros were building their reputations. Adam Tooze, the Columbia historian and author of the Chartbook newsletter, recently tracked what has changed . For much of the 2000s, foreign official buyers—reserve managers in export-oriented economies—absorbed a significant share of new Treasury issuance. Countries running trade surpluses with the United States accumulated dollars and recycled them into government debt. That mechanism has weakened substantially since the global financial crisis, and particularly since 2020. In its place, domestic and foreign private investors—including, prominently, hedge funds operating through offshore financial centers such as the Cayman Islands—have become the marginal buyers of U.S. government debt. “The Cayman islands matter,” Tooze wrote, “because they are the offshore home for a significant cluster of hedge funds. And since the 2010s it is hedge funds who have provided a key source of demand for US government debt.” He noted that Bloomberg’s Tracy Alloway has charted the rise of private investors in the Treasury market, as seen below in purple. The irony is not subtle. The global-macro industry that Druckenmiller helped build—the one that made its name by betting against governments—now finances the government whose fiscal credibility it once tested. And the instruments have changed along with the players. Where Druckenmiller’s generation took outright directional positions against currencies and interest rates, today’s hedge funds increasingly participate in the Treasury market through basis trades: exploiting the spread between cash bonds and futures contracts using significant leverage. Tooze cited a New York Fed analysis estimating that hedge funds held $2.4 trillion in long Treasury exposure as of September 2025—exceeding holdings by mutual funds and U.S.-chartered depository institutions. The 50 funds with the largest gross Treasury exposures accounted for about 90% of the total. Aggregate basis-trade volume stood at roughly $830 billion, close to twice its early-2020 peak. The precedent for what happens when that capital moves quickly is March 2020, when a rapid unwind of leveraged positions contributed to a breakdown of Treasury-market liquidity severe enough to require Federal Reserve intervention. The exposures are now substantially larger. The macro irony This puts Druckenmiller’s injunction to “let the bond market speak” in a more complicated light. His warning is first a fiscal one: the long bond reflects inflation, growth expectations, fiscal supply and confidence in the government’s willingness to confront its deficits. Dulling that signal, he argues, only delays the political confrontation needed to reduce the primary deficit. “If the 30-year must trade at 5.5% to clear,” he wrote in a somewhat obvious voice that’s become synonymous with the use of AI, “that isn’t a crisis. It is an invoice.” But market prices also reflect market structure. A move in long-term yields can be a referendum on fiscal credibility, a reflection of inflation expectations, or a product of the mechanics by which leveraged positions are financed, hedged and distributed. Often it is all three. If Treasury reacts to ordinary yield increases as though they were an emergency, it may make investors doubt its commitment to price discovery. If it disregards genuine strains in the fragile architecture that now absorbs so much government debt, it risks allowing a liquidity problem to become a disorderly unwind. The task for Bessent is to demonstrate that Treasury knows the difference. But there is an irony in the setup that Druckenmiller himself didn’t acknowledge: the investors now pressing on long-term yields are, structurally, the same kind of actor Bessent spent his career being. The situation “has echoes of the past,” Hilsenrath said, adding that the past is never a perfect framework for thinking about the present. The Bank of England in 1992 was trying to defend a price on its currency that the market had found indefensible, he said, whereas in this case, Bessent is trying to defend a price in the Treasury market that the market found indefensible. Beyond a currency and a Treasury bond being somewhat apples and oranges, he added, “the problem might be, we shall see, more profound in the sense that the market seems to be waking up to the idea that U.S. fiscal policy is unsustainable.” That’s the question Druckenmiller is now raising, he added: whether the United States is willing to confront a fiscal position that its own creditors are beginning to doubt. When asked about the use of AI in writing the essay, Hilsenrath said he didn’t especially care, because he thought Druckenmiller made several “brilliant” points and, ultimately, he put his name on it. “The good news is that if you get serious about addressing these issues, you can actually fix them,” Hilsenrath said. “The problem—as anyone in Spain, Italy or Greece can tell you—is that if you let the market impose a fix on you, it’s going to be a lot more painful.” For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing. This story was originally featured on Fortune.com

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Luna, 26, no longer speaks to her father. There was no single betrayal, no dramatic rupture. What ended the relationship was his insistence on obedience and dominance. “In his head it’s like, ‘I’m the father, you’re the daughters, we do things my way,’” she told me. Luna explained to him that she was “not willing to have a relationship like that,” and her terms for contact were simple: “We could only have a relationship where you understand that I’m a person, and I’m not just gonna do whatever you want me to do whenever you want me to.” He refused to accept those terms. So after numerous attempts to make it work, she walked away. Family estrangement has become a fixture of talk shows , advice columns and social media . It’s also the subject of my new book, “ Families We Lose .” Much of the current media coverage on estrangement tends to frame it as an epidemic that is the product of therapy culture . My research suggests that this framing misses what is actually happening. I conducted 68 in-depth interviews with adults who went no-contact with a parent, sibling, grandparent or other relative. (All participant names, including Luna’s, are pseudonyms used to protect their privacy.) I found that estrangement is less a story of trauma or selfishness, and more a clash between competing expectations: an insistence on loyalty, duty and forgiveness on one hand, and the longing for respect, emotional safety and personal growth on the other. Falling birth rates raise the stakes Estrangement is fairly common. In a 2022 study , my colleagues and I used national survey data that followed parent-child relationships over several decades. In that period, 26% of adult children in the U.S. experienced a period of estrangement from their fathers, and 6% reported a period of estrangement with their mothers. Our findings were consistent with what sociologist Karl Pillemer found in his 2020 national survey , which showed that 27% of Americans were estranged from a relative, or roughly 67 million people. Headlines often treat numbers like these as evidence of a sudden crisis. But no study has tracked estrangement rates over time, and nothing in the existing data shows that it is skyrocketing. Yet estrangement can have different consequences today. People have far fewer children than in most of human history , so the loss of one relationship may carry greater weight. Two generations ago, a parent might have had several other children with whom to maintain close ties; today, many parents have only one or two. Smartphones, meanwhile, have created expectations of constant contact and connection within families . Even adult children who have quietly distanced themselves by moving away or limiting contact may still find themselves fielding regular requests for connection through texts, phone calls and social media. This can be stressful, emotionally taxing and make it difficult to establish boundaries. Those I interviewed often had difficult childhoods, and some had experienced abuse. But they rarely pointed to childhood as the proximate cause of the estrangement, instead citing the poor quality of their relationships with parents and other relatives in adulthood. ‘Kinship culture clash’ For them, the path to estrangement was usually gradual, marked by mounting tensions and unresolved conflicts. It emerged out of the recognition that a parent or sibling would not, or could not, treat them the way they desired. For most, this looked like accountability and growth, intentional connection, mutual respect and safety. Molly, 36, described a lack of accountability as motivation for her estrangement from her mom, Cindy, a decade ago. Molly told Cindy: “You’ve done all these horrible things. I just need you to acknowledge it. I feel like I’m owed an apology.” Molly felt that her desire for accountability was a basic need of any relationship, noting, “I didn’t feel like that was asking too much, but apparently it was. I put the ball back in her court, and she’s just sitting on it.” To what extent are family ties unconditional? Natalija Grigel/iStock via Getty Images Most of the people I interviewed recognized that they had different definitions of what family should look like and feel like. In my book, I call this a “kinship culture clash.” Divorce offers a familiar example of this kind of shift. For most of American history, marriage was an institution shaped by economics, law, religion and social pressure . You stayed married because that was what one did, and there were few alternatives that didn’t leave one spouse – usually the woman – socially ostracized and financially destitute. Then, as sociologists Andrew Cherlin, Anthony Giddens and Stephanie Coontz have argued , marriage transformed into a relationship judged by its ability to provide love and support, promote ideals like fairness and lead to personal growth. Once a marriage started being thought of as an arrangement that needed to work for each spouse, divorce became thinkable, then ordinary. Should family ties always bind? The family you are born into can carry the same tension. Ties to parents and siblings are often expected to be permanent and unconditional, reinforced by cultural norms and laws that bind parents and children to one another, even when children become adults. For example, in adulthood, 27 states have filial piety laws that require adult children to bear financial and, in some cases, direct caregiving responsibility for aging parents. This is part of what I term “compulsory kinship.” Under this framework, you show up at Thanksgiving no matter what was said last year, because your family is your family. You take your dad into your home to care for him when he gets sick, because he’s your father. This model does help many Americans. It can ensure aging parents have adult children to care for them, while adult children may have a place to call “home,” no questions asked. In contrast, the estranged people I talked to for my book adhere to what I call “democratized kinship,” and it generally reflects views of contemporary marriage. Family bonds require mutual respect, reciprocity and support in order to continue. Under this model, genetics take a back seat: Anyone can become family, just like anyone can become a spouse, but no one is automatically family. Neither family framework is new, and neither belongs to a single generation. Some people have always held family of origin relatives to distinct standards of care. Queer people have been building chosen families for generations. But what I chart in this book is the conflict between these two ways of thinking. Estrangement often happens when these two ideas of family collide and cannot be reconciled – when one member insists that biological or origin ties and hierarchies reinforce the family unit, and another insists that behavior does. That is Luna’s story. Her father drew on the logic of age and hierarchy – “Because I’m your father and I said so” – and she demanded to be treated as a person first and a daughter second. A family that works for you People who leave their families of origin are not rejecting family altogether. In a recent study drawing on these same interviews, I show that estrangement operates as a catalyst for what I call “redoing family.” Adults who cut off their relatives can still create kinship networks. Marina Khromova/iStock via Getty Images After cutting off contact with a family member or multiple relatives, people commonly went on to rebuild kinship networks, drawing friends, partners, in-laws and neighbors into roles once reserved for blood relatives, and constructing new family units organized around support and reciprocity. Luna is not on her own because she no longer talks to her father. Her family today includes her sister, whom she talks to multiple times a day, and a large group of friends. “I think you can have your chosen family, and whoever in your biological family you connect to,” she told me. “Everybody just gets to choose who they feel they have a real, intimate, deep connection with. And that’s your family.” She didn’t want less family. She simply wanted a family that works. Rin Reczek , Professor of Sociology, The Ohio State University This article is republished from The Conversation under a Creative Commons license. Read the original article . This story was originally featured on Fortune.com

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Finance Minister Enoch Godongwana has argued that despite the National Treasury’s desire to widen the net of social assistance as much as possible, it “simply cannot afford to do more.” This counters the Institute for Economic Justice’s (IEJ) argument that the government can and should do more regarding the threshold value and the grant’s value. The finance minister, Social Development minister and the South African Social Security Agency (Sassa) appeared before the Supreme Court of Appeal on Tuesday to appeal a judgment of the Gauteng High Court in January 2025 that some of the Sassa regulations limit access to the R370 social relief of distress (SRD) grant, and are therefore unconstitutional and invalid. The focus is on the online applications, which the IEJ argues restrict the poor from applying for the grant, and the definition of “income”, which it argues is too broad because it could include financial support received through other means. The High Court agreed, declaring that the “income” in the regulation means money received on a regular basis from formal or informal employment, business activities or investments. “It is declared that the words ‘financial support’ in Regulation 1 means money received on a regular basis which benefits the recipient, that does not constitute income, and which the recipient has a legal right to receive,” reads the judgment. ‘Misconception’ Advocate Gilbert Marcus, for the finance minister argued that the IEJ’s argument that the government can do more to reach more beneficiaries by increasing taxes or borrowing more money is based on “a misconception of both the nature and enforcement of socio-economic rights and the separation of powers.” “The attack on the various regulations and its acceptance by the High Court must be approached in the context of the extraordinary complexities of the competing demands on the state’s resources,” argued Marcus. “The High Court dismissed concerns in this regard without a proper evaluation of the evidence and without a proper appreciation of the limits of judicial intervention in complex matters of budgeting and the need to balance a multitude of competing claims.” He argued that the High Court weighed in on the national budgeting process, ignoring the institutionally equipped body to decide how public revenue should be spent more effectively. Poverty Advocate Marcus further argued that evidence from the Treasury established a concerted effort to address poverty. “Treasury is acutely conscious of extreme levels of poverty faced by too many South Africans. And it has an appreciation which lies, as they put it, at the root of much of its approach to the budget as a whole,” argued Marcus. This, he said, is evident in the government’s expenditure on the social wage. Between 2014 and 2027, South Africa will have spent between 55% and 61% of its budget on it. This includes social security, education, health and housing. “And as is explained, government can only expend money it extracts through taxation. Borrowing money is nothing more than deferred taxation,” argued Marcus. “And the demands of the state concern virtually every facet of society, including obligations flowing from the socioeconomic rights in the constitution, access to adequate housing, access to health care services, sufficient food and water. The list literally goes on and on. And all this in the context of a relatively small tax base, with tax evasion and non-compliance. “Treasury says that further or continued expenditure on the SRD grant, by definition, must come at the expense of some other expenditure.” Marcus highlighted Treasury’s long-term plan to promote economic growth and create jobs, a more sustainable solution. Grants affordability The IEJ has argued that the grant amounts have not been progressive either, to keep up with inflation. It argues that the grant’s value today is lower than it was in 2020. However, Marcus said the notion that the government can afford to do more by raising taxes or borrowing is “artificial and unsustainable”. “That argument simply ignores the evidence which establishes in the supplementary affidavit from Treasury that when the R20 increase was mandated, the country’s financial position had deteriorated further. The increase meant that the state had to cut other expenditure to give at least some further relief,” argued Marcus. Marcus said the relief sought by the IEJ would require additional expenditure between R93 billion and R139 billion. This in addition to the government’s concession that around 18 million people may be eligible to receive the R370 grant, while it can only budget for around 10 million. ‘Grants are all the same’ Further, it is the IEJ’s argument that all grants are the same and should be treated the same with regard to the threshold value and the grant amount; Marcus said this argument was incorrect. “The SRD grant was introduced as a temporary measure under the Disaster Management Act, which by definition was intended to have a limited life only. “Under the Disaster Management Act, there are specific periods for which disasters can be declared, and measures taken pursuant to a disaster have the lifespan of the disaster alone. “SRD grants are legislatively treated differently from the other grants. The Act, in section 1, defines ‘social grant’ and ‘social relief from distress’ separately. “The definition of social grant does not include social relief from distress, and whereas the social grants as defined are prescribed in Sections 6 to 12 of the Act, social relief from distress is regulated by Section 13, which leaves it to the Minister to prescribe by regulation.” Online applications The minister of Social Development and Sassa argued against the attack on the online application process, saying it was the most successful process to facilitate the SRD grants. They argued that the SRD grant was temporary, as a recipient’s economic circumstances could change at any moment. Hence, the government requires a monthly verification process to ensure the grants go to qualifying recipients. The online application and monthly verification process is the only effective way to ensure the grant reaches the right recipients, they argued. Remedy Marcus said, although the finance minister stands by all his arguments, should the IEJ succeed, four months would not be sufficient to implement the changes. “The court knows more than enough to understand the complexity of the issue. The order of the High Court is extraordinarily prescriptive, and if that order is to stand, there will be questions of compliance thereafter.

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Mumbai, August 25, 2026: The BMC's crackdown on unauthorised alterations at Rustomjee Crown, a three-basements-plus-63-floors high-rise luxury residential complex at Crown Mill Compound in Prabhadevi, continued for the second consecutive day on Tuesday. On Monday, demolition of encroached areas of flats in B wing was undertaken and on Tuesday, demolition was under way on the 23rd and 38th floors of A wing, sources said. BMC Issues Notices It was after complaints of unauthorised construction and FSI violations, followed by notices by the BMC to the owner/occupier under Section 53 (1) of the MRTP Act, that the civic body started the demolition of encroached areas. "A total of 138 notices have been issued to the owner over the last six months. The other party was given sufficient time and hearings to present their side and all Supreme Court guidelines were followed before starting the demolition action of unauthorised areas," said a senior BMC officer. Rustomjee crown tower at Prabhadevi The unauthorised alterations in the building include encroachments like merging of the common lobby into the habitable area of the flat, which are not in the BMC's approved plan. The BMC's speaking order dated 10/3/2026 to the owner/occupier of the respective flats of A and B wings of Rustomjee Crown, where unauthorised alterations were made, states that a notice was served under the MRTP Act on 24/7/2025 directing the noticee to submit the approvals within 30 days and, further, a personal hearing was also held on 25/2/2026, when the owner/occupier was directed to get the said work regularised or restore it. However, in the second inspection by the civic officials, it was observed that the notice requisitions were not complied with, the speaking order adds. Nine Alterations Removed As per the BMC statement issued on Tuesday night, in the two-day action, nine unauthorised constructions/alterations were removed. The unauthorised alterations included a common lobby illegally enclosed using wooden materials and lightweight Siporex bricks and merged into the habitable area of the flat, a portion of the kitchen duct illegally enclosed using M.S. plates and merged into the kitchen area, part of the ventilation shaft included in the flat area using M.S. plate and used as a storeroom, part of the powder room included in the kitchen area and others. Developer Denies Violations Clarifying on the demolition action by the BMC, a company spokesperson, Rustomjee Group, said, "The construction was undertaken in accordance with the plans sanctioned and approved by the competent authorities as per 1991 DCR. It complies with all the necessary norms and regulations and has received Occupation Certificate about two years ago. There are no violations on part of the developer. We have been fully cooperating with the authorities and have provided all necessary details and information sought by them. We have also been engaging with the members of the building and encouraging them to undertake the necessary corrective measures, wherever required." Meanwhile, activist Santosh Daundkar, who had filed a complaint under 'Public Grievances' to Additional Municipal Commissioner (City) on November 17, 2025, alleged that a blind eye was turned by DMC Zone II, Chief Engineer DP and Chief Fire Officer to open violations brought to their notice about unauthorised encroachment carried out of the common lobby area into a habitable room at Rustomjee Crown (Phase I), Prabhadevi, and has demanded an investigation by the Economic Offences Wing (EOW) and the Enforcement Directorate (ED) into the financial transactions associated with this project and wrongdoings in the project. Also Watch: Mumbai: BMC Demolishes Unauthorised Alterations In 3 Rustomjee Crown Flats Over Common Area Encroachment "The luxury towers have been constructed on land meant for mill workers, and that several public representatives and other individuals have acquired properties in the project. The flats in this project are reportedly valued at crores of rupees and considering the scale and value of the transactions, there is a strong suspicion that financial transactions exceeding 250 crore, including possible unaccounted or non-transparent transactions, may have taken place in connection with the project. Appropriate legal action should be taken against anyone found responsible for wrongdoing," Daundkar said. To get details on exclusive and budget-friendly property deals in Mumbai & surrounding regions, do visit: https://budgetproperties.in/

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

问题是这样，我们有 2 组服务器，A 组和 B 组，一部分用户从 A 组下载文件，一部分从 B 组下载文件。分成 2 组的初衷是，B 组用户下载量通常比较大，并且是离线服务，意味着可以慢一些，但不能影响 A 组用户。现在的问题是，A 组 和 B 组最终下载的来源都是 S3，依然存在共享资源，会出现问题：在下载量比较大的时候，占用全部的 S3 上传带宽，导致 A 组用户收到影响。 我们希望限制 B 到 S3 的带宽，比如 10Gbps. 有几个限制： 我们无法在 Client – Server 这里限流，因为 Server 侧有 Cache，我们希望 Client – Server 这里依然可以使用非常大的带宽来下载； S3 的服务器没有限流的功能，这些服务器我们也无法控制，我们只能配置 Server A 和 Server B； 这个问题的难点在于：流量整形一般都是做在 egress 的，因为 egress 天然有一个 queue，发送端可以决定在什么时候发出去 packet，从而把发送速率可以稳定控制在一个想要的比例。 但是我们这里的场景是 ingress，对于 ingress 是没有 queue 的，流量整形一般也无法作用在 ingress ——因为包已经被物理网卡收到，DMA 完成，作为 ingress 侧是无法决定这个包什么时候到达的，只能决定怎么处理这个包。所以 ingress 侧一般是 classification, action，即 drop。 基于 Drop 的 policing（真形象的名字，超速开罚单的警察） 就有一个问题：由于 TCP 的拥塞控制机制，发送的带宽会逐渐上涨，最终会超过 10Gb。限流开始生效时，sender 瞬时发送速率超过 10 Gbit/s，并持续消耗 policer 的 token；当 token 不足时，后续 packet 被判定为 exceed 并丢弃。 大量包被丢弃导致 TCP 的发送端认为发生了拥塞，会大幅下降 cwnd。实际测试，如果用 policing，实际能跑的速度大概在 5-7Gb，而且不稳定。 以下是一个模拟环境，使用 policing 限速在 1Gbps，其 throughput 如下图所示： 可以看到限速在 1Gbps 的时候，实际速度在 850Mbps 左右震荡 Window Scaling 图 再看 window scaling，可以看到 outstanding bytes 的上包络线呈现出明显的 AIMD/CUBIC loss-response 特征；结合 sender 侧的 ss -ti 可以确认 cwnd 在 loss 后也发生了类似下降——在上升到一定的大小之后，就会因为超过 policing 的限速而被 drop 一些，TCP 在发现有丢包之后，就降低 cwnd，从而降低发送速度，符合 AIMD。（这个环境用的拥塞算法是 cubic，默认的参数 cat /sys/module/tcp_cubic/parameters/beta 是 717，下降的比例是 717/1024 ≈ 0.7，和图中相符，图中是从 500kb 下降到 350kb）。 这样用户就不高兴了：为什么实际只能用到 5Gb 左右，不是说好了 10Gb 的吗？ 如果要做到规整的 10Gb，这里就需要用到流量整形了，traffic shaping。 Linux 有一个特殊的驱动：ifb，全称是 Intermediate Functional Block ，是一个虚拟网络设备，主要用来把原本的 ingress 流量重定向到一个“可做 egress qdisc”的设备上，从而实现 ingress shaping。 本质上，就是在收包链路上加一个网卡绕一下，这个网卡会调用 egress 的代码，这样我们就可以使用 tc 原来的逻辑做 egress 的 traffic shaping 了。 加上 ifb 虚拟网卡之后，包的路线如下： IFB 加入之后的包路线 在 ifb_xmit() 的时候，看起来是一个负责发送包的代码，实际上它的逻辑是：修改 skb 的 dev 为原来真实的进来的那个 dev 1 ，（而不是 ifb 这个虚拟的），然后进入到 netif_receive_skb() kernel 网络栈里面。 这样，我们就可以用这个 egress 的 qdisc 做 shaping 了。 实现的代码如下： 第一步，创建 ifb 虚拟接口，并且把流量从物理网卡 redirect 到 ifb： modprobe ifb ip link add ifb0 type ifb ip link set ifb0 up tc qdisc add dev bond0 handle ffff: ingress tc filter add dev bond0 parent ffff: protocol ip u32 \ match u32 0 0 \ action mirred egress redirect dev ifb0 第二步，创建一个 HTB，有 2 个 class，一个有限速 10Gb，另一个没有，可以跑到物理线速。 tc qdisc add dev ifb0 root handle 1: htb default 20 tc class add dev ifb0 parent 1: classid 1:1 \ htb rate 50gbit ceil 50gbit tc class add dev ifb0 parent 1:1 classid 1:10 \ htb rate 10gbit ceil 10gbit tc class add dev ifb0 parent 1:1 classid 1:20 \ htb rate 50gbit ceil 50gbit 第三步，把要限速的 IP 放到 1:10 class： tc filter add dev ifb0 parent 1: protocol ip prio 1 u32 \ match ip src $ip/32 \ flowid 1:10 然后测试吞吐，证明实际的 throughput 稳定在 956 Mbits/sec。 在 ifb 参与的 traffic shaping 下，限速可以稳定在 950 Mbits cwnd 也在稳定地上涨 这样就达到了我们的目的：这条链路的速度稳定地使用限速 1Gbps 左右。 有几个小问题值得讨论。 为什么是 950 Mbits/sec 而不是 1Gbits/sec？ tc 做限制的时候，计算的是 skb 的 size，1500 Bytes，而这张图（以及使用的 iperf3 的显示）是计算的 TCP 的 body，即 MSS，1448 Bytes。1000 × 1448 / 1500 ≈ 965 Mbit/s。 在 cwnd 的图里面，刚开始的时候有一些丢包，为什么？ TCP 启动时通常先经过 slow start，cwnd 快速增长；进入 congestion avoidance 后，CUBIC 按三次函数继续探测可用带宽。发生 loss 后，它记录此前的窗口位置并降低 cwnd，随后重新逼近旧的 Wmax。 在这个图里面，只有一开始的时候有 loss，后续几乎没有 loss，贴着 throughput 发，说明 traffic shaping 做的不错。 按照 TCP 的拥塞控制算法，AIMD，即使是 traffic shaping 不也应该在最大窗口附近震荡吗？为什么这个图看起来这么平滑？ 一个原因是，在逐渐逼近 1Gbps 的时候，limit 已经不是 cwnd 了，这种情况下 cubic 不会再激进地增加 cwnd 2 。通过 ss -ti 可以看到，大部分的时间卡在 sndbuf_limited:4688ms(29.6%) 。 ESTAB 0 3646016 [::ffff:10.10.10.2]:5201 [::ffff:10.10.10.1]:59444 cubic wscale:13,13 rto:228 rtt:24.801/0.155 ato:40 mss:1448 pmtu:1500 rcvmss:536 advmss:1448 cwnd:2918 ssthresh:1717 bytes_sent:1893410080 bytes_acked:1889764624 byt es_received:37 segs_out:1307809 segs_in:652610 data_segs_out:1307808 data_segs_in:1 send 1.36Gbps lastrcv:15848 pacing_rate 1.64Gbps delivery_rate 961Mbps delivered:1305291 app_limited busy:15848ms rwnd_limited:12ms(0.1%) sndbuf_limited:4688ms(29.6%) unacked:2518 rcv_space:14600 rcv_ssthresh:42230 notsent:560 minrtt:3.024 如果调整一下参数，扩大 wmem，cwnd 实际也会震荡的。 最后一列是 cwnd，也在不断震荡 最后一个问题也是最关键的问题，为什么在 traffic shaping 之后，cwnd 的变化速度变得这么平滑了呢？都是一样的 cubic 算法，为什么不像 policing 那样剧烈下降？以及为什么这里 send buffer 出现了瓶颈？ 使用 policing 时，包进入 ingress 只有 2 中 action： 接受，然后 kernel 的网络栈会 ack，处理包； 直接 drop，没有 ack，导致 cwnd 下降。 使用 ifb 的 traffic shaping 之后，ingress 的包进入物理网卡，被 mirror 到 ifb，然后在这里经过 queue 排队。经过流量整形之后，包到达 kernel 的网络栈，然后发出 ack。这里的关键是，tc 添加了一个漏斗，大量的包（当然，buffer 需要内存的，内存是有限的，如果大到一定的程度就只能 drop 了）进来之后都存起来，然后用固定的流速放出去。只有被放出去的包才进入网络栈，只有进入了网络栈，kernel 才会给 sender 发回去 ack，此时，sender 的 cwnd 才有了新的空闲，新的包得以发送。 再换一个角度解释一下，从 sender 的角度，假设 cwnd 是 4，sender 发送了 4 个包，在收到 ACK 之前，无法再发送新的包了。这时候，receiver 的 shaper 每秒通过一个包，即每秒会发回来一个 ack，这样，后续的每一秒，比如5，6，7，8，sender 都可以再发送一个包到网络上。 send buffer 是另一层限制。TCP 已经发送但尚未被 ACK 的数据仍然需要保留在 send buffer 中，以便发生丢包时重传。因此 IFB shaping 不仅延迟 ACK、限制 cwnd 中空间的释放，也会延迟 TCP send buffer 中已发送数据的释放。当 send buffer 不够大时，sender 甚至可能在 cwnd 尚未用满之前就受到 send buffer 限制。 所以 IFB 以固定速率释放数据后，ACK 也以相应的节奏返回。ACK 一方面释放 cwnd 中的 outstanding 空间，另一方面使已经确认的数据可以从 send buffer 中清除。两者共同使 sender 的发送节奏逐渐跟随 shaper 的速率。 即，网络上最多可以飞多少，是 cwnd/rwnd 等窗口决定；什么时候可以继续注入新 packet，则由 ACK clock 驱动。（而 ACK 也会让总的 cwnd 提高）。 以上就叫做 TCP 的 self-clocking 3 ，ACK 本身就像“时钟脉冲”，会决定发送速度。 这个机制有一个关键就是包没有被丢弃，只是 ack 的时间晚了一些，如果包被丢了（比如 policing），那么 cwnd 会下降，无法匀速发送。 https://github.com/torvalds/linux/blob/master/drivers/net/ifb.c#L115 ︎ if (!tcp_is_cwnd_limited(sk)) return; https://github.com/torvalds/linux/blob/master/net/ipv4/tcp_cubic.c?utm_source=chatgpt.com#L326 ︎ 在 Computer Networking: A Top-Down Approach Chapter 3.7 也有介绍 ︎

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Short answer: use an emailed, single-use reset link as the default for most SaaS login recovery, and add SMS OTP only where users may genuinely lack email access or the product already maintains verified phone numbers. Email is usually the simpler system because the login identifier, recovery destination, and support workflow can remain in one channel. SMS can shorten the interaction, but it adds phone-number lifecycle, message segmentation, regional consent, and delivery-state work. “Cheaper” depends on your traffic and failure rates, so model completed recoveries rather than message sends. This is a recovery decision, not a notification preference. The goal is to return the right person to an account without turning a delayed message, an expired credential, or a recycled phone number into an account takeover or a support queue. I've worked around enough spam filtering, rate limiting, and OTP delivery gaps to treat the channel as one component of that system — never as the system itself. What should a SaaS password recovery flow use: email API or SMS OTP? Start with the account data you can already trust. If every user signs in with an email address and changing that address is a controlled operation, an email reset link creates the smaller data surface. The service generates a high-entropy, single-use token, stores only a protected representation of it, sends a link, and accepts that token once before a short expiry. The browser then moves the user into a password-change session. An SMS OTP flow looks compact on screen, yet the backend has more questions to answer. Was the phone number verified recently? Can the user update it without being signed in? How are country codes normalized? What happens when a number is reassigned? Does the support team have a safe path for a person who lost the device? A six-digit form doesn't make those policy decisions disappear. So the default is straightforward. Choose email first when email is the stable account identifier and recovery is occasional. Consider SMS as an additional path when the service has a legitimate reason to collect and continuously verify phone numbers, or when an email-only path would strand a meaningful user group. Don't add SMS merely because entering a short code feels faster in a demo. The catch is that email is not suitable when users routinely lose access to the mailbox that identifies them. In that case, a separately verified recovery factor or a support-assisted process may be necessary. SMS is also not suitable when phone ownership is weak evidence in the product's environment, when regional messaging operations are not staffed, or when collecting phone numbers would create disproportionate privacy and compliance work. Neither channel is a universal fallback. The delivery constraint comes before the screen For email, delivery begins with domain alignment and message integrity. DKIM lets a signing domain take responsibility for a message and lets a receiver detect changes to signed content in transit. That is useful infrastructure, but a valid signature is not a promise that the message will reach the inbox or that the sender is trustworthy. Recovery design still needs stable sending identity, conservative content, bounce handling, suppression, and monitoring around the entire path. Password-reset mail should also be boring. Keep the purpose obvious, avoid unnecessary tracking, don't include the current password, and make expiration clear without exposing account state. The request endpoint should give the same public response for an existing and a nonexistent account. Otherwise the recovery form becomes an email-address discovery tool. SMS has a different physical constraint: encoding changes capacity. A single GSM-7 message can contain up to 160 characters, while UCS-2 reduces that to 70. Concatenated messages reserve characters for segmentation, leaving 153 GSM-7 or 67 UCS-2 characters per segment. A curly quote, non-Latin name, or translated sentence can therefore turn one planned message into multiple segments. That affects cost and makes truncation tests important. Tiny changes matter. Keep an OTP message short, put the code near the start, and don't ask the user to reply. Test the exact production template with every supported locale and with the encoding calculation used by the messaging path. Your mileage may vary by destination and carrier; delivery telemetry from the actual launch countries is what resolves that uncertainty, not an estimate made from an English template. Channel delivery is asynchronous in both cases. The UI should never imply that “sent” means “received,” and a retry must not silently create several simultaneously valid credentials. This is where otherwise tidy implementations tend to fray: each click creates another token, delayed messages arrive out of order, and the user enters a code that was valid two requests ago. Make recovery one state machine, not two endpoint piles Model email links and texted codes as two presentations of the same recovery challenge. A challenge has a subject, a channel, a destination snapshot, a creation time, an expiry, an attempt count, a send count, and a terminal state. It should reveal as little as possible in logs. In particular, log a challenge identifier and transition reason, not the raw token or OTP. The core transitions are small enough to review: From Event To Required behavior absent recovery requested pending Return a neutral response and apply abuse controls pending delivery accepted active Start or retain the same challenge validity window active valid proof submitted consumed Issue a narrow password-change session active invalid proof submitted active or locked Count attempts without revealing account existence active expiry reached expired Reject the proof and require a new challenge any nonterminal state newer challenge wins superseded Prevent old and delayed messages from succeeding That last row deserves attention. “Newest challenge wins” is easy for a user to understand, but it requires atomic invalidation. Another valid policy is to keep one challenge and resend the same proof within a bounded window. Pick one policy, document it, and test concurrency. Mixing the two creates hard-to-reproduce failures. Here is a deliberately channel-neutral core. It omits storage and transport details so the security-sensitive decisions remain visible: from dataclasses import dataclass from datetime import datetime from enum import Enum class Status(str, Enum): ACTIVE = "active" CONSUMED = "consumed" EXPIRED = "expired" LOCKED = "locked" @dataclass class RecoveryChallenge: challenge_id: str proof_digest: bytes expires_at: datetime attempts: int max_attempts: int status: Status def verify_challenge( challenge: RecoveryChallenge, candidate_digest: bytes, now: datetime, ) -> Status: if challenge.status is not Status.ACTIVE: return challenge.status if now >= challenge.expires_at: return Status.EXPIRED if challenge.attempts >= challenge.max_attempts: return Status.LOCKED if candidate_digest != challenge.proof_digest: challenge.attempts += 1 return Status.LOCKED if challenge.attempts >= challenge.max_attempts else Status.ACTIVE return Status.CONSUMED In production, compare secret-derived values with a constant-time primitive, make the consume transition atomic, and bind the resulting session to password change only. A successful proof shouldn't mint a normal long-lived login session. Also decide what happens to existing sessions after the password changes; that is an account policy, not an email or SMS detail. Abuse controls belong around challenge creation and verification. Rate-limit by several signals rather than one obvious address, because a single-key limit either blocks households and offices or remains easy to distribute around. The public response stays neutral while internal metrics distinguish suppressed, attempted, accepted, delivered, expired, locked, and consumed states. Those distinctions are how an operator finds a delivery gap without leaking it to an attacker. Compare completed recovery cost, not message price A per-message quote cannot answer which channel is cheaper. Use a simple cost model that includes sends, retries, engineering ownership, compliance work, support contacts, and abuse. Then divide by successful recoveries. A channel with a lower send cost can still be the expensive choice if delivery failures or confusing retries create more support work. The comparison should be made per region and per user cohort: Decision factor Email reset link SMS OTP Existing account data Often already present for SaaS login Requires a collected and verified phone number Message sizing Template size is rarely the billing unit Encoding and concatenation can change segment count Recovery interaction Open link, then set password Copy or autofill code, then set password Delivery operations Domain signing, bounce and suppression handling Country, carrier, encoding, consent and number lifecycle handling Lost destination Mailbox recovery or support path Device loss, number change or reassignment path Best fit Email-address accounts with infrequent recovery Products already operating verified mobile identity Do not average the US and EU into one line on a spreadsheet. Languages change SMS encoding and template length. Consent, retention, sender identity, and support procedures also need review in each launch market. I'm not sure any generic traffic estimate can settle the decision for a particular SaaS product; a small, instrumented rollout with real destination mix and a written compliance review can. There is a DX cost too. Two channels mean two template systems, two sets of delivery events, and twice the temptation to let behavior drift. Keep one challenge API behind them and make transport adapters translate provider-specific delivery states into a small internal vocabulary. That preserves the option to change transports without rewriting account recovery policy. Roll out the least complex path first Ship the state machine and email path behind a feature flag, with synthetic checks that request a recovery, retrieve the test message, consume the token, and confirm that reuse fails. Add dashboards for time to delivery, time to consume, resend rate, expiry rate, lock rate, and support contacts. Avoid publishing a single “delivery rate” that hides whether messages were merely accepted upstream. Then test the ugly edges: concurrent requests, an expired link in an old tab, a password change while another recovery is pending, casing and Unicode in identifiers, delayed delivery, and a user who requests several messages quickly. Exercise deployment rollback with challenges created by both application versions. Recovery data outlives a request and can cross a release boundary. Add SMS only after writing down the user cohort it rescues and the operational owner for it. Reuse the same challenge rules, validate each localized template's segment count, and stage availability by country. Compare completed-recovery and support outcomes against the email path; don't interpret raw send volume as success. The simplest channel is the one your team can operate correctly on a bad day. For the usual email-identified SaaS account, that is an email reset link. SMS earns its place when it solves a documented access problem and the product is prepared to own phone-number and regional messaging constraints. References RFC 6376: DomainKeys Identified Mail (DKIM) SMS character limits and segmentation (GSM-7/UCS-2)

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

IT之家 8 月 25 日消息，苹果今日在新款 Mac mini 中推出了 M6 芯片。M6 是苹果首款采用 2 纳米制程的尖端芯片，全面升级了所有计算单元，使性能在各个维度均得到提升。该芯片配备了一个拥有全球最快 CPU 核心的 12 核 CPU、一个集成神经加速器的 12 核 GPU、一个双 16 核神经引擎，以及 170 GB/s 的统一内存带宽。 M6 专为日常用户、学生、开发者、AI 爱好者及企业用户的工作流设计，可平衡性能、能效与端侧 AI 能力，让用户能够轻松应对日常任务、编程及创意项目。 M6 采用最先进的 2 纳米制程工艺，通过在更小的芯片面积内集成更高的晶体管密度，实现了性能与能效的显著飞跃。M6 还引入了双 16 核神经引擎，其峰值计算能力可达前代产品的 2 倍，从而大幅提升设备端 AI 工作流的运行速度。系统框架可自动同时调用这两款引擎，使应用程序能够获得更快的模型执行速度。 M6 搭载了全新的 12 核 CPU 架构，比 M5 多出两个核心，该架构由 2 个超级核心、4 个性能核心和 6 个能效核心组成。与 M5 相比，M6 提供了全球最快的单线程性能，多线程性能最高可提升 1.2 倍；与 M1 相比，其多线程性能最高可提升 2.4 倍。2 个超级核心专用于处理单线程工作负载，性能核心功耗更低，可与超级核心协同运行高负载的多线程工作负载，而能效核心则负责处理日常后台任务。苹果表示，所有这些功能均具备行业领先的能效比。因此，诸如图像编辑、代码编译、新文件索引以及运行智能体 AI 工作负载等高负载 CPU 任务的执行速度均比以往更快。 M6 搭载了 12 核 GPU（比 M5 多 2 个核心），且每个核心均配备神经网络加速器。与 M5 相比，该设计使 AI 场景下的 GPU 峰值计算性能提升了近 30%；与 M1 相比则提升了 8 倍以上，从而在与设备端大语言模型（LLM）交互时实现显著更快的提示词处理速度。 此外，M6 还搭载了苹果最新的先进图形处理能力，包括对着色器核心架构的升级、动态缓存技术以及硬件加速光线追踪功能。 多种技术相结合，可以为游戏带来更加出色的视觉真实感、更快的渲染速度以及更高的帧率。M6 在处理复杂图形时，其几何顶点处理率提升了 50%。 M6 支持最高 32 GB 的统一内存，可支持在资源密集型应用间进行多任务处理，并在设备上运行大型语言模型（LLM）以执行安全、私密的智能体任务。此外，它还提供高达 170 GB/s 的统一内存带宽，较 M5 提升了 10%，较 M1 提升了 2.5 倍。 苹果表示，其开发者框架与工具 —— 包括 Core AI、Core ML、Metal 及 Xcode 可直接调用 M6 芯片的先进硬件资源。开发者可充分利用 M6 芯片中的双 16 核神经引擎实现卓越的 AI 计算性能。借助这些开发框架和新型芯片，开发者可在 Mac 设备上本地运行并微调大型 AI 模型。苹果的开发者工具和框架可自动优化 CPU、GPU 及神经引擎的性能表现，并使开发者能够利用 Apple 基础模型或其自主研发的专有 AI 模型，在设备端构建并运行强大的 AI 工作负载。

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

#独立开发者 🧱 做 App 订阅页没参考？这里整理了 1110 个真实案例 Open Paywall Gallery 是 PaywallPro 公开的 iOS 订阅 App 付费墙与 Onboarding 数据集。 目前收录了 1110 个 App，可以查看真实付费墙截图、Onboarding 预览、定价模型、付费墙模式和部分变现数据信号。 适合独立开发者用于参考做付费转化页面 📮 投稿 📢 频道 💬 吹水 🌐 网站

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

此前高通曾确认，2026骁龙峰会将于当地时间9月22日至24日正式举行，全新一代骁龙8系旗舰平台将会在这次活动上亮相。现在随着时间的推进，更多新品相关爆料也陆续出现。 最近的一份消息显示，型号为 V2606A 的 vivo 手机现身了 GeekBench 跑分库，且搭载了 SM8975 芯片。结合跑分列表来看，这颗 SM8975 芯片采用 2+3+3 集群 CPU 设计，主频峰值超过 5.0GHz。在 GeekBench 6.7.1 基准下，单核成绩 3434 分，多核成绩 9334 分。 结合博主@数码闲聊站 的爆料来看，SM8950的命名为Snapdragon 8 Elite Gen6，即第六代骁龙8至尊版；SM8975命名为Snapdragon 8 Elite Extreme Gen6，即第六代骁龙8超级至尊版。也就是说，最新曝光的跑分成绩属于第六代骁龙8超级至尊版。 而关于这一代骁龙8系旗舰芯片，以往也出现过不少相关爆料。同一位博主的爆料曾提到过，SM8975基于2nm，2+3+3新一代Oryan CPU架构，共享16MB L2，频率更激进，A850 GPU 18MB GMEM，支持LPDDR6/5X，峰值性能刷新高；SM8950基于2nm，2+3+3新一代Oryan CPU架构，共享16MB L2，A845 GPU 12MB GMEM，支持LPDDR5X，兼顾一定性能与功耗。以往的消息显示高通测试了 6 个第六代骁龙8超级至尊版芯片样品，但该芯片上线后仅有支持 LPDDR5X 的普通版和顶配 LPDDR6 版两个版本。 消息显示，6 款芯片样品，大致可以按照内存标准划分为 LPDDR6（后缀 AC）和 LPDDR5X（后缀 BC）两大类，顶级旗舰机型可能配合使用 LPDDR6 版本与 UFS 5.0 存储，而注重成本的厂商则可通过选择 LPDDR5X 版本大幅节省开支。还有消息称，这 6 种配置通过调整不同版本 CPU 和 GPU 的时钟频率，高通能以更低价格出售性能略有缩减的芯片版本。同时，所有六种测试配置均采用统一板号，并完整支持 sub-6GHz 和毫米波 5G 网络。 据悉，第六代骁龙8超级至尊版芯片是安卓最强 2nm 芯片，主要在 2027 年发布的 Ultra 旗舰手机中进行搭载，但与此同时其价格也会更高一些。基于此也有消息显示，年底的旗舰机会区分档位，好几家只有Pro Max级别的机型才能用上顶配满血SoC。 以往的爆料显示，高通第六代骁龙 8 至尊版芯片将分为标准版与定位更高的版本两款。其中，定位更高的版本定价预计将突破 300 美元。而与此同时，第六代骁龙 8 至尊标准版芯片的价格有很大的可能是不会上涨的。这也就是意味着，接下来应该会有不少旗舰产品会选择搭载这颗芯片。 近期文章精选： 玄戒O3发布！小米18 Fold正面对决苹果折叠 小米18标准版12月发！18 Pro升级细节来了 苹果下月发布会前瞻，或推8款新品 华为阔直板手机公布！这个比例 感受下？ 雷军确认新玄戒芯片将至！小米神秘新机9月发 文章原文 平台地址： http://www.jintiankansha.me/t/MOXYnCAUta

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

Dolly's Tennessean Travel Stop is so much more than a place to get fuel and a quick snack. Renée Reizman I visited Dolly Parton's Tennessean Travel Stop, a fairly new spot in Cornersville, off of I-65. It had a dog park, fuel stations, several Dolly Parton-themed restaurants , and road-trip essentials. The exterior and bathrooms were also beautiful. This was a great place to stop on a long road trip. On a rainy day in rural Tennessee, I began seeing billboards for Dolly's Tennessean Travel Stop off of I-65. I'd been on a long, cross-country road trip with my dog and the weather had placed me behind schedule. This stop, located in Cornersville, Tennessee, would be a welcome break. I was already planning to visit Dollywood, Dolly Parton's theme park in the Smoky Mountains, but when I learned the Queen of Country had opened this attraction of sorts in late June, I planned a detour to check it out. As a veteran road tripper, I was curious to see how it compared to other beloved roadside chains like Buc-ee's, Wawa, and Love's. The stop seemed pretty promising from the outside. 005 Renée Reizman The travel stop is about an hour south of Nashville and open 24/7 — it seemed quite busy when I arrived around 5:30 p.m. on a Saturday. The truck stop looked quite large at around 25,000 square feet on 18 acres, though the parking lot felt smaller than I expected. After seeing the Buc-ee's in Luling, Texas — the largest convenience store in the world, spanning almost 80,000 square feet — I was imagining a space that could fit a couple of football fields. This lot, however, wasn't much bigger than your average QuikTrip's. Still, l easily located an empty spot. I could already hear live country music wafting from the establishment, which got me hyped. It's not just pet-friendly — there's also a dog park. Renée Reizman Before even entering the building, there was a lot to see. My first priority was taking my road-trip partner, my Chow Chow, Flora, to Doggy Parton. This modest-sized, fenced-in dog park was on par with the ones at most Love's truck stops, but it had much more personality. The fence was adorned with silhouettes of Parton juggling a tangle of leashes, getting pulled by what appeared to be Great Danes and German shepherds. It was a hit with my pup. Renée Reizman The grassy park looked very clean and had multiple waste-bag dispensers, trash cans, and even a picnic table. Flora loved that she could be off-leash after so many hours in the car. The truck stop looked vibrant and photo-worthy. 004 Renée Reizman After the dog park, I had to indulge in a couple of outdoor photo opportunities. One of Parton's old tour buses was parked beside a colorful mural depicting Tennessee's pastoral, rolling hills. The side of the blue bus had a giant retro photograph of Parton behind the wheel, her name written in rhinestones, and a winking "DLY 925" license plate. A rainbow music staff created a walkway from the bus to the front entrance. On the side of the building was the mural, "Butterfly Lullaby (Mountain Gospel)," designed by the artist Britt Flood. The giant butterfly in its center attracted many other travelers, who were queued up to snap photos as they grew wings. Nearby, I spotted an outdoor patio with lawn games and more Instagram potential between the cute retro pickup truck with flowers spilling out of the flatbed and another large mural. The stop has practical offerings for road trippers. 026 Renée Reizman Finally, I stepped foot inside the travel stop. I entered through the far left entrance, which is closer to the trucker's fuel pumps. It brought me into perhaps the most practical part of the establishment. Right away, the beverage dispensers and shelves of food reminded me of a typical convenience store but more elevated with shiny, dark tiles and walls decorated with vinyls and images of Parton. Inside, there are plenty of grab-and-go snacks and essentials. 025 Renée Reizman Like any gas station, there was a fridge of cold drinks, large carafes of coffee, road-trip snacks (like beef jerky and gummy worms), and plenty of automotive accessories to help a long-hauler. There are showers and laundry available, too. 009 Renée Reizman I was glad to see that the stop wasn't just playing with trucking as a gimmicky aesthetic to entertain tourists, but was actually providing valuable services to professionals passing through town. There was a lounge for truckers decorated with tour posters, showers available to use for $16 (or free with a fuel purchase), and a laundry room charging $4 a wash. Beyond that, there is a ton of merchandise. 011 Renée Reizman Past this section of gas-station staples, the building opened up to a full Parton paradise. An enormous chandelier made up of sparkling guitars hung over rows of bedazzled merchandise. I held back on buying too much, but later regretted it. Renée Reizman Knowing that I'd be at Dollywood in a matter of days, I restrained myself from stocking up on a sparkling blue travel mug or a Jolene fanny pack. I only grabbed a souvenir magnet for a steep $14.95. Later, I regretted my self-control because many of these products weren't available at the theme park. I especially wished I had splurged on a bright-pink tank top embellished with the glittering mudflap-girl-style Parton for $29.95. Next stop was the restrooms at Dolly's. Renée Reizman Once I wove through the merch, I headed to the restroom. Popular Southern rest stop Buc-ee's has a reputation for having incredibly clean bathrooms, which is one of the most welcome sights during a long drive. So, my bar was pretty high for Dolly's. Even the bathrooms here felt photo-worthy. 014 Renée Reizman Given the stop's popularity, I thought there could have been more stalls, but the facilities looked sparkling clean. The bathroom also just looked beautiful. It was covered with large black-and-white portraits of Dolly getting ready in her dressing room, the perfect backdrop for a bathroom selfie. In terms of food and beverage options, there was a whole Dolly Parton universe. 012 Renée Reizman I had three options for dining: the grab-and-go DLY BBQ, an unnamed full-service restaurant that offered a Southern-style buffet for $21.99 per adult and barbecue plates, and Cup of Ambition for coffee, drinks, pastries, and soft-serve ice cream . There was also a bar above the restaurant, stylized with dark glam and a large portrait of Parton from her "Rockstar" album era. (At the time of my visit, it seemed it had been reserved for a private event.) For dinner, I headed to the main restaurant, where there was live music being performed. Renée Reizman The large restaurant had high ceilings, a large wall of windows, and a stage — not a huge surprise considering live entertainment is also part of the usual lineup at Dolly's. Typically, there's live music every Friday, Saturday, and Sunday. While I waited to get seated, I caught a band performing a country-style cover of The Killers' popular hit "Mr. Brightside." The restaurant also had a large buffet with instruments displayed above it. 017 Renée Reizman From the line, I could also see the buffet, which had a salad bar and lots of Southern fare, like mac and cheese and fried chicken. Above it, there were banjos, guitars, and other colorful instruments on display. Buffets have never appealed much to me, so I instead planned to order the meat and two sides dinner for $15.99. After waiting for about 20 minutes, I decided to grab something else. 020 Renée Reizman Unfortunately, luck wasn't on my side. There were several large parties ahead of me, so I gave up waiting for a table after about 20 minutes. I still had another hour to drive in the dark, after all. Instead, I decided to grab a quick and convenient bite from DLY BBQ. My dinner to-go was pretty decent. 029 Renée Reizman I got a $9.99 brisket sandwich from the grab-and-go area at DLY BBQ. I didn't think the sandwich was as good as the one I've had with sliced brisket at Buc-ee's for around the same price, but it was still moist and tender, and outranked most gas-station food . I also noticed that a famous Dollywood treat was available for purchase. 013 Renée Reizman Though I passed on dessert, I appreciated that both DLY BBQ and Cup of Ambition sold loaves of Dollywood's famous cinnamon bread for $12.99. I couldn't believe that people could now get the coveted confection without having to make a trip to Pigeon Forge . I grabbed a refresher from Cup of Ambition before I left. 023 Renée Reizman Cup of Ambition had quite the beverage menu, including things like refreshers, tea, milkshakes, and espresso-based drinks made with Community coffee. There was even an option to add a "Dolly Dollup" of whipped cream to any iced drink for $1. It came in flavors like banana puddin', chocolate cinnamon bread, caramel, and strawberry. I settled on a strawberry-acai refresher, and the regular size (my only option besides large) cost $4.95. The caffeinated drink was sweet and tasted very similar to a Starbucks Energy Refresher . There was so much to see that I almost forgot to fuel up. 007 Renée Reizman I enjoyed Dolly's Tennessean Travel Stop so much that I nearly forgot to get gas on my way out. The pumps use gas supplied by Exxon, and the setup looks no different than a typical station. Prices felt pretty on par with what I'd seen nearby, too. I made sure to top off my tank before heading back onto the wet roads to complete the day's long drive. All in all, I'd totally come back. 024 Renée Reizman Compared to popular rest stops like Buc-ee's, Dolly's totally holds its own. This was worth the detour, and even though it was a bit crowded, I'd make sure to stop again if I were near this leg of the I-65. It's definitely a fun place to stop if you need to refuel, grab refreshments, or just take a break with some live music. Read the original article on Business Insider

为什么现在还能做

基于本期筛选信号整理

建议怎么做

建议结合原文验证并小步测试

变现说明

可通过产品化服务、订阅或咨询变现

可复制性

具备明确需求后可进一步复制

机会视角：可围绕该信号探索可复制的产品或服务

信号

小游戏市场日耗1.8亿但产品生存艰难，AI将研发周期从30天缩至3-4天，素材产量成为门槛，竞争转向流量经营和商业化闭环。

为什么现在还能做

2026年下半年，微信和抖音策略分化：微信重留存和跨端，抖音重内容和跨场景（短视频/小说/短剧）。

建议怎么做

建立系统化买量思维：结合内容获客（短视频/直播，抖音直播投放增750%）、社交获客和搜索获客；采用混合变现（IAA+IAP）；注重跨平台适配而非简单移植。

变现说明

通过降低综合获客成本、提升用户LTV（生命周期价值）来实现盈利；混合变现模式能扩大用户底盘并挖掘付费深度。

可复制性

需要团队具备研发、素材、直播、投放、商业化的综合闭环能力，分散的团队弯路更多；适合有运营经验的中小团队。

机会视角：小游戏不再是暴利快钱，而是精细活。AI让生产变容易，但用户变难搞了。能活着的是那些把获客、留存、变现串联成系统的人。

“普通量化：人写规则 → 静态系统；我的量化：自己积累经验 → 动态成长。第1周按基础逻辑判断，第4周积累几百条经验后遇到熟悉场景能直接调取，第3个月胜率开始指数级提升。”

观点增量

将AI Agent的决策过程与历史经验库绑定，通过复盘归档（成功/失败案例+场景标签）实现自我迭代，这是构建智能副业工具（如自动交易、内容推荐）的核心架构思路。

机会视角：别只做一次性工具，要做能“越用越聪明”的系统。每一次错误都是喂给未来的数据，这才是AI副业的护城河。