Yesterday, I used @dappOS_com's xBubble Coding to build a small tool specifically for monitoring top account interactions and event market signals. The reason is simple. I saw a top exchange founder reply to a content account's tweet and even liked it. I checked the related token's K-line chart, and the market cap immediately surged by tens of millions! If you rely solely on manually scrolling through tweets, it's hard to catch these opportunities early. Especially for on-chain memes, many attention signals don't start from announcements. It could be a like from a top account. Maybe a reply from the exchange founder. Or an interaction between the project team and a KOL. At that moment, I thought, if such signals could be automatically captured and then connected to a strategy for automatic buying, wouldn't that be insanely profitable! Seeing the news before others and selling early to make a profit. Coincidentally, I recently saw an official xBubble demo featuring a Polymarket event detection and trading assistance case, which is very similar in concept—not manually monitoring every piece of information, but linking events, signals, prices, and market changes into a single workflow. Following this idea, I used xBubble Coding to build a Web3 trading signal tool. Test version: The process roughly goes like this: when a new interaction signal appears on Twitter, the system automatically captures it first. After capturing, it passes through a strategy matching layer; only if it hits the rules I set does it proceed. Then it checks a whitelist database to filter out the assets I truly care about. Next, it checks wallet cache for related addresses and historical behavior. Then it generates a token price briefing. For event markets like Polymarket, it also synchronizes event status, odds changes, related information sources, and potential trading directions. Finally, it submits the qualified content to a preset on-chain workflow. This is the real result of "tracking top accounts" and "tracking event markets." It turns a light interaction signal into an executable trading observation chain. Previously, after spotting related signals, I manually checked the coin, looked at prices, browsed wallets, and checked on-chain data. By then, it was already time to take the loss. Now, signals enter the process themselves, and the system helps me complete filtering, organizing, and next steps first. This change is crucial. Because the real value of xBubble Coding is not just generating a piece of code for me, but packaging a specific scenario into an end-to-end business SOP. What I made this time is more like a small Web3 trading information service. The front end monitors top account interactions and event market detection. The middle handles strategy matching, asset filtering, and information source organization. The back end connects price briefings, wallet cache, odds changes, and on-chain workflows. Once this chain runs smoothly, it’s not just a tool but a personal information advantage system that can continuously iterate. This is also where I see xBubble’s business value. It’s not selling code itself but end-to-end business services. For small teams and individuals, what really holds people back is often not "can I generate a page," but "can I link content, data, processes, payments, execution, and follow-up modifications into a closed loop." xBubble’s SOP system core is lowering this threshold. You can use it to build a World Cup merchandise store, quickly linking product pages, materials, payments, and order processes. Or create a traffic rules learning product, connecting content flow and private domain conversion. Or build a local service portal, letting users view services, submit requests, and complete payments. Or like me, build a Web3 event detection, top account signal tracking, and Polymarket trading assistance tool. These needs share a common trait: very vertical, very niche, very close to real business. They don’t necessarily require a large tech team but need a set of execution processes that can be quickly launched, continuously modified, and directly receive revenue. This is exactly where xBubble fits OPC. One person or a small team, as long as they know who they serve, what signals to capture, what products to sell, and how to deliver, can use SOP to build the business. More importantly, xBubble is not limited to fixed templates. Mature scenarios can quickly start with ready-made SOPs, while new niche needs can continue to generate more tailored SOPs through the Bubble Engine. This is especially important for Web3. Because opportunities change fast here. Today it might be top account interactions. Tomorrow Polymarket event changes. The day after, some wallet behavior. The next round could be community points, airdrop tasks, or on-chain data dashboards. If the tool can only handle one fixed scenario, the space is very limited. But if SOPs can keep generating with new scenarios, individuals and small teams can continuously distill their niche judgments into new business processes. My setup for top account and Polymarket event monitoring is a small but very concrete example. It turns my usual actions of scrolling Twitter, watching the market, checking coins, wallets, and event odds into an automatically running process. In the future, I want to add new account sources, I can expand. Want to change strategy matching rules, I can adjust. Want to refine briefing fields, I can modify. Want to connect more on-chain actions, I can continue extending. This is what AI tools in the OPC era should really look like. Not forcing individuals to imitate big companies building teams, but enabling individuals to gain near-professional execution ability through SOP. Automatic content generation, multi-platform promotion, stablecoin settlement, on-chain task execution, client onboarding, backend management—these steps that originally required multiple collaborators are compressed into a lighter business system. So now when I look at xBubble Coding, I no longer see it just as an AI Coding tool. It’s more like foundational business startup infrastructure for solo entrepreneurs and small teams. For me, top account interactions are no longer just information points I handle after seeing them. Polymarket event changes are no longer just pages I remember to check. They can all become strategy inputs that automatically enter the system, automatically filter, organize, and trigger follow-up processes. Such niche needs were rarely addressed before because they are too personalized, too vertical, and change fast. But xBubble’s SOP approach is just right for turning these needs into low-cost verifiable, continuously iterated, and even service-solidified things. This might be its most valuable aspect. The first half of AI Coding is enabling more people to create applications. xBubble wants to do the second half: make these applications truly run business, generate revenue, and continuously iterate commercial processes. For OPC, this is far more important than a pretty demo.

When I just saw this post, I was replying on my iPhone. After reading it, I glanced down at the logo on the back, feeling quite conflicted. Apple's most important manufacturing partner in India, Tata Electronics, was hacked by a ransomware group called World Leaks. On June 12, these people uploaded over 200,000 files stolen from Tata onto the dark web, totaling 630GB. Tata only publicly admitted it on June 22. For a full ten days in between, the data was openly available on the dark web, and no one said a word. According to Reuters, the leaked content includes the supplier list for the iPhone 18 Pro, detailed parts lists, and actual photos of drop tests conducted at Tata's factory, dated earlier this year. This group had previously stolen data from Dell last year and Nike earlier this year. Why do I feel conflicted? Because I hold memory stocks, and the list of Apple’s memory suppliers is included in this leak. Recently, Apple raised prices on Macs and iPads by 15% to 25% due to memory price hikes, and now the details of who supplies memory and how much have been fully exposed. For the suppliers named, the damage from this list being revealed could be even greater than the price increases themselves. Ironically, Apple has been aggressively shifting production capacity from China to India in recent years to diversify risk. India now produces 26% of the world’s iPhones, up from just 6% four years ago. But just as the eggs were moved to a new basket, that basket got completely wiped out. Leaked details about the 18 Pro from various sources: the A20 Pro chip has a new packaging, new colors including dark cherry red, the Dynamic Island is nearly halved in size, the front camera is upgraded to 24MP, and the Pro Max battery might exceed 5200mAh. Three models may launch simultaneously in September: 18 Pro, Pro Max, and the foldable iPhone Ultra with a 7.8-inch inner screen, priced above $2000. The standard 18 model is delayed until next spring.

Can a person's foot really beat Buddha's foot?? 😂 Cape Verde may not lose, but Argentina may not win either. This year's World Cup is really interesting. A bold prediction: Argentina is definitely stronger, but the knockout stage probably won't start with them completely dominating; the pace is likely to be steady. Cape Verde got this far not by luck, but by defensive discipline and counterattack efficiency. As long as they can withstand Argentina's ball control pressure in the first half, this match has a chance to be dragged into a tight contest. The scenario could be Argentina scoring first through individual skill, and Cape Verde equalizing in the second half via set pieces or counterattacks. The 90 minutes end 1-1, then go into extra time, which would be a scoreline with an upset flavor but not absurd. Australia 1-2 Egypt Colombia 2-1 Ghana @bcgame

In late June, SK Hynix once surpassed Samsung to become the most valuable company in South Korea, with its stock price reaching 2,987,000 KRW, nearly hitting 3 million KRW per share, and the screen was full of "Storage is King." Today, the stock plummeted; Hynix closed down 14.57%, Samsung down 9.06%, and the KOSPI fell 7.89%. The Korea Exchange also triggered a sell-side sidecar, halting programmatic sell orders for five minutes. There are two triggers. One is the rumor that Meta plans to build cloud services to rent out its unused AI computing power; the other is a report from The Information that OpenAI has improved inference efficiency, reducing some inference costs. Together, these sparked the market’s biggest fear: if AI can squeeze more from existing hardware, is the demand for new chips less urgent? Has computing power been overbuilt? Overnight, U.S. chip stocks already took a hit: Micron dropped over 10%, the Philadelphia Semiconductor Index fell over 6%; Japan’s Kioxia also dropped over 13% today. Hynix is just the one that stood the highest and fell the hardest in this wave. But looking deeper, the simplest reason is that it had risen too much. The gains this year have been extremely exaggerated, priced by the market as nearly perfect. For such stocks, the biggest fear is someone questioning if it’s really that perfect. The news from Meta and OpenAI provided exactly that excuse. Has the fundamental collapsed? Not at the moment. While the stock plunged today, some brokerages raised their target prices; IBK raised Hynix’s target to 4 million KRW and expects it may beat expectations for the 11th consecutive quarter, which aligns with Micron’s recent strong earnings report. The drop today seems more like a correction of the price inflated by sentiment; the business itself hasn’t taken that hit yet. Among these news items, one I truly believe is AI computing power overcapacity. Right now, it’s just a trigger for sentiment, but it points to a real issue. The entire industry is investing heavily in building computing power and stockpiling memory; if one day it turns out they built too much and demand doesn’t keep up, the strongest foundation of the memory business will really start to loosen. My small position is also in the red today, and it’s painful. But I won’t move it because of a single profit-taking candle driven by sentiment. Is there a memory shortage today? Yes, there is. When there isn’t, I’ll come back and tell you.

Xiaohongshu is about to ring the bell on the Hong Kong Stock Exchange, and this news has been trending recently. But honestly, I’m not that interested in the bell-ringing itself. What really caught my attention was a new feature quietly launched before the IPO called REDSkill. I write about US stocks and IPOs every day, so I’m very familiar with the storytelling before a company goes public. SPCX told their story this way, OpenAI did the same, and many companies before listing first set valuation anchors and filled the imagination space. You have to make the secondary market believe that your current valuation isn’t the end—you have another layer of story. So when I first saw Xiaohongshu talking about REDSkill, my initial reaction was: is this just another new narrative packaged for the IPO? A former Xiaohongshu employee once bluntly said that because they hadn’t gone public yet, Xiaohongshu needed a new story even more. That sounded like a complaint, but it’s actually quite accurate. Xiaohongshu definitely needs a new story; just relying on the phrase “grass-planting community” (a term for product recommendation community) is something the market is already very familiar with and hard to imagine much beyond. To go public, they have to tell investors that they’re not just a content community—they can grow something new. But this time, I think it’s a bit different. Because REDSkill isn’t just a “future plan,” it’s something you can actually open and use. Creators can make Skills that can be directly attached under notes. When you come across a piece of content, you can try it out right away—no need to jump to another page, no downloads, no technical knowledge required. This is a key point. Many platforms have made tools before, but they just put the tools there waiting for you to find them. Xiaohongshu this time is different—they embed Skills into the content stream, placing them in the same context as notes, likes, favorites, and comments. In other words, they’re not selling a tool; they’re making the tool part of the content. We can already see some early data. Some notes have driven over 10,000 uses of a single Skill; there’s a World Cup anime wallpaper Skill that reached 1.5 million exposures; and REDSkill-related notes have directly surpassed 10,000 likes. These numbers aren’t extraordinary for a mature product, but the catch is that it’s still in internal testing. That’s interesting because Xiaohongshu’s strongest suit has never been tools but distribution. As long as something can be packaged by content, planted by users, and repeatedly spread by the community, it has a chance to create a flywheel effect. Alibaba, Tencent, and ByteDance are also pursuing similar directions but with different approaches. Some focus more on selling capabilities, some on selling services, and some want to integrate into mini-program ecosystems. Xiaohongshu’s path leans more toward content community: first let users see it, then let users easily use it, and finally let users spread it themselves. This is what I’m truly paying attention to. In the mobile internet era, the App Store is the gateway. Whoever controls the gateway controls users’ time and commercial flows. If we really enter the era of Agents and Skills next, then Skills could very well be the new generation of apps. Whoever can distribute these Skills will have a chance to grab the next gateway ticket. Xiaohongshu is probably trying to grab that ticket now. Of course, I don’t want to overstate things. REDSkill is still very early, and internal test data can’t be directly equated with ecosystem success. Whether it can continuously attract creators, keep ordinary users engaged long-term, and truly commercialize remains to be seen. But at least this time, it’s not just a pure concept. It’s not just talking about what they want to do in the future; there’s already a group of users actually using it, notes generating data, and a visible chain of dissemination. This is quite rare among many pre-IPO stories I’ve seen. #redskill #rednote #skill #Xiaohongshu

The headline about the return of a veteran public blockchain—over the past few years, I've seen at least eighty if not a hundred of these. Basically, the community releases a roadmap, posts a partnership poster, there's a buzz for three days, then it falls silent again. I'm too familiar with this pattern, so when I clicked on TWIN's one-year summary these past couple of days, I honestly didn't have high expectations. But! This summary contains no roadmap. It lists real countries, real ports, real goods. On the Africa route, TLIP processed over 180,000 commercial invoices and more than 300,000 declarations; at the UK border, a pilot was done with over 2,000 batches of frozen poultry transported from Poland, with data delivered up to twenty hours in advance to regulators; export documents from Kenya to the UK used to take eight hours by air and three weeks by sea, now they can be obtained within five minutes. The UK government even assigned four cabinet office members to work full-time on this for an entire year. Moreover, its ambition goes beyond this. At the end of last year, @iota partnered with the African Continental Free Trade Area Secretariat, the World Economic Forum, and the Tony Blair Institute to launch a program called ADAPT, built directly on TWIN. The goal is to connect all 55 African countries into a single open digital trade infrastructure by 2035, doubling intra-African trade and unlocking over $70 billion in additional trade volume. The first step was taken this May, with Kenya, Morocco, and Nigeria becoming the first three countries to implement it, covering East, North, and West Africa respectively. This is no longer a pilot project; it’s a continent in motion. I stared at these numbers for a long time. Because this isn’t about future trade being changed by blockchain—it’s a record of change already happening. A single cross-border trade can involve thirty stakeholders, thirty-six documents, and over two hundred paper copies. This mess isn’t some minor Web3 issue; it’s a decades-old mountain of problems in global trade that hasn’t been cleaned up. What IOTA is doing now is making the blockchain the layer of trust and audit in trade, not just launching another on-chain app that no one remembers after a hundred days. TWIN has only been online for a year, yet it has already moved from a concept to a real pilot spanning Africa and the UK. If this speed sounds slow in Web3, put it in the context of global trade and you’ll see how difficult it is! It requires negotiating with governments, customs, ports, exporters—slowly piecing together a network that everyone can use but no single party controls, from processes that no one wants to relinquish control over first. Pushing this to implementation within a year speaks volumes. In a bull market, everyone competes to tell the most polished story; in a bear market, you see who’s quietly getting things done. Too many projects on the market tell stories they themselves can’t clearly say where they will land, but @iota over the past ten years seems to have slowly turned that story into the flowers of East Africa, the poultry of Poland, and the $400 a month saved by an exporter in Kenya.

Fun fact! Now, every 43 seconds, a new .ai domain is registered on Earth. Who ends up with the money? It's hard to guess. It's a small island in the Caribbean called Anguilla, with a population of 15,000—fewer people than many towns on the mainland. Back in 1995, when the internet was just starting, countries were assigned two-letter suffixes: China got .cn, the UK got .uk, and Anguilla was assigned .ai. At that time, these two letters were almost worthless; no one imagined that thirty years later, they would become the most valuable two letters in the world. In 2021, this island earned just over $70,000 a year from .ai domains, which was less than a Silicon Valley engineer's two-month salary. Then ChatGPT came along. The first thing startups did was grab a .ai domain to appear connected to AI. AI companies worldwide started sending money to this small island. Every time a domain was registered, the treasury rang a bell. In 2023, this income surged to over 80 million East Caribbean dollars—more than 200 million RMB. By 2025, domain registrations accounted for nearly 40% of the government's regular revenue. For every 10 dollars the government earns, 4 come from these two letters. Moreover, .ai domains require continuous renewal, and fees from registrations, renewals, and transfers keep flowing steadily into Anguilla's treasury. Many people think wealth comes from hard work, but sometimes it can just fall from the sky. Anguilla's case is like winning by inheriting a prosperous legacy 😂