---
title: "Two People Can Build a Global Business: AI Opens a New Route Overseas for Chinese Entrepreneurs"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2025-12-05T06:59:33+00:00"
canonical: "https://ffcap.cn/en/research/src-20251205-01html"
source: "https://uniqueresearch.substack.com/p/src-20251205-01html"
language: "en"
---

# Two People Can Build a Global Business: AI Opens a New Route Overseas for Chinese Entrepreneurs

_Original · Unique Research · 2025-12-05_

Historical edition: This preserves the article and roundtable as reported in December 2025. User counts, revenue and conversion examples, launch timing, product uniqueness claims, forecasts and views on marketing workflows belong to the original report and named speakers, not independently verified present-day results. The Imagen 3 identification is the source editor’s conjecture, not a confirmed model name. Boolean Vector and Wilderness Stars below are descriptive English renderings of 布尔向量 and 旷野群星, not verified official English company names. Where the source gives an amount without a currency, no currency is inferred.

If your team had only two or three people—or even just you—would you dare imagine doing business worldwide rather than running a stall in Shenzhen's Huaqiangbei?

At a roundtable titled "E-commerce and Going Global: AI Strengthens Individual Entrepreneurs' Global Competitiveness" during the 2025 Beijing Unique Awards, five guests examined the question thoroughly. The moderator was Wang Chaochao, partner at Unique Capital. The five guests were Wei Lihua, founder of global language-social platform HelloTalk; Wang Qing, founder and CEO of Boolean Vector, which builds cross-border e-commerce advertising-creative Agents; Shu Junliang, CTO of MuleRun, described as a Taobao for AI Agents; Peng Yunye, COO of QuickCEP, which develops intelligent cross-border customer service; and Liu Rushan, founder and CEO of Wilderness Stars, which lets users build video-production pipelines with natural language.

They broke down how AI has already changed the thresholds, paths, and moats for individuals expanding overseas.

This article follows their conversation from the tool boom back to people and business themselves.

I. Why Would a Small Team Dare Do Business Worldwide? The Answer Begins with Language

In the past, language formed a wall in front of Chinese individual entrepreneurs.

If you wanted to build a product for the world, localization alone was enough to deter most people at the outset. Websites, in-app copy, email marketing, and customer-service replies all had to be translated into accurate and natural Japanese, Korean, Spanish, and other languages, with context and cultural differences also considered.

For an individual or small company, this was almost an impossible task.

HelloTalk founder Wei Lihua offered a direct comparison from his own experience.

Previously, serving users across dozens of languages required an entire localization team. Today, AI performs most of the same work.

Even if you can understand only a little English, AI can generate natural multilingual versions of webpages, advertising copy, and email content without requiring you to find one translator in every country.

More importantly, even a product such as HelloTalk, which already serves more than 60 million users worldwide, uses AI translation extensively rather than merely treating AI as a toy to try.

When leading companies begin using AI as infrastructure, it is no longer a trend but the new normal.

Once language is no longer a barrier, the nature of going global changes. The old logic was, "Your English is poor, so do not even think about expanding overseas." The new question is, "Can you explain clearly what you want to do?" AI can handle the rest.

II. From Exporting Goods to Exporting Agents: What Is Being Sold Has Also Changed

When people once spoke of going global, the default image was small commodities from Yiwu, 3C electronics from Shenzhen, and factories in Dongguan, with containers leaving ports for destinations around the world.

AI is now quietly rewriting that picture.

MuleRun CTO Shu Junliang is building something easy to visualize: put simply, it is a Taobao or App Store for AI Agents.

On the platform, an Agent is itself a type of merchandise that can be placed on a shelf and sold to users worldwide.

He described a representative case: one day, a major overseas company released a new image-generation model.

Within the first few hours after the model's release, an individual developer on MuleRun realized perceptively that it was becoming a hot topic.

In roughly one hour, using an extremely simple Prompt, he created an Agent with a straightforward core function: upload a hand-drawn sketch and generate a rendered image with one click.

The Agent was not complex, had no sophisticated interface, and could hardly be said to possess a technical moat.

But it entered at the correct moment: the model had just launched, the market was still excited, and users wanted the simplest possible way to experience it.

The outcome was highly concrete. Within one week, the individual developer earned more than US$1,000 from that small Agent.

The boundaries of going global are expanding:

Previously, going global meant sending Chinese-manufactured goods abroad;

Today, it can mean packaging your Prompt, workflow, or domain Know-how into an Agent and selling it to users worldwide.

Containers may still sail across the sea, but invisible trade routes are also appearing in browsers.

For individuals, this path offers a clear advantage:

You need neither a factory nor a warehouse—only a solution capable of completing a closed loop and a small group of real users willing to pay for it.

III. Video Pipelines and Influencer Marketing: How Did Six Months Become Half a Day?

If eliminating the language barrier gives more people an opportunity to go abroad, changing the mode of content production turns global expansion from a game of chance into industrial production.

Liu Rushan's product Mulan.pro uses an interesting phrase: it moves AI video production from handicraft into an assembly line. In the past, a cross-border seller conducting influencer marketing followed roughly this process:

First, find an influencer, often arranging meetings across time zones;

Then send samples and wait for delivery and testing;

Then communicate repeatedly about the script, filming, and editing while adapting to the creator's own pace.

The complete cycle lasted at least several weeks and sometimes several months.

After TikTok e-commerce exploded, that pace was clearly no longer adequate. The platform changes too quickly and traffic windows vanish instantly. While your influencer video is still being filmed, a competitor may already have completed three rounds of A/B testing.

AI video workflows change this tempo. The current approach is:

A seller uses a workflow to generate complete video content, with the setting, actions, captions, and pace all highly controllable.

Then the seller asks the influencer one question: "I will swap your face into this template. Will you post it?"

The other party agrees, and the swapped video is posted. A process that once took six months is compressed into half a day.

More impressively, the workflow can continuously produce videos that feel familiar without being duplicates.

The structure can remain consistent—for example, the same opening demonstration, feature highlights, usage scenario, and concluding Call-to-Action section.

Yet the person, angle, actions, and lines vary slightly in each video, avoiding the platform's duplicate-content algorithm while giving users the sense that the store is updating diligently.

For a team of two or three people, the scariest problem is not a lack of creativity, but a lack of time.

The central value of a workflow is to turn continuous creation from repeated late nights into one-click execution.

When you have a personal AI production studio that can replicate output almost infinitely, the question is no longer whether you can create, but what exactly you want it to express.

IV. Where Should AI Investment Go? Do Not Confuse a Demo with an Outcome or Search for Nails While Holding a Hammer

Return to the most practical question facing individual entrepreneurs:

With limited time, energy, and capital, where exactly should you invest in AI?

1\. Insight and Sales Are the Two Ends Always Worth Heavy Investment

Wei Lihua's answer is clear: first use AI for two purposes—market insight and product validation, and marketing and promotion.

For example, use AI to map the competitive landscape for a specialized product quickly, or let foundation models scan Instagram and TikTok to identify potential competing products, popular content, and what users complain about in comments.

Alternatively, use AI to automate email marketing and generate multilingual email content at scale, or have AI collect the follower lists of a particular category of blogger to support more precise outreach.

In cross-border e-commerce, this combination of insight and outreach is no longer merely conceptual.

According to the report, QuickCEP's customer data demonstrates one point: the more expensive and complex a product, the higher the conversion rate when AI provides shopping guidance.

When buyers make high-value decisions, they need detailed, patient answers available at any time, not a human customer-service representative who appears enthusiastic but is actually very busy.

Here, AI customer service does not replace people. It compensates for humans' natural limitations involving time differences and patience.

2\. Outcome Orientation: Deploy AI Heavily Only Where It Actually Makes Money

Wang Qing of Boolean Vector poured some cold water on the discussion from another angle.

Many people believe AI can do everything, but in his view, it has not been deployed unless the outcome can be quantified using ROAS (return on ad spend).

For example, the first 3 seconds are the most important part of a short-video advertisement.

If AI selects unusable footage while screening creative assets, it is helping you waste money rather than earn it.

There are two lessons here:

First, do not blindly trust the dazzling moments in foundation-model Demos.

Many demonstrations look as if they are only 1% away from perfection. When you build a real product, however, you may discover that the remaining 1% conceals another 70% of engineering work.

Individual entrepreneurs who mistake a Demo for an outcome can quickly collapse both in their schedules and psychologically.

Second, understand exactly what you are buying.

You are not buying a collection of AI features, but higher conversion, lower customer-acquisition cost, and faster experimentation.

If those metrics do not change, the AI system is merely an attractive shell.

3\. Do Not Build AI + X; Build X + AI

Shu Junliang's advice directly exposes a common blind spot.

Over the past two years, many people have begun by asking: "Can I build AI + e-commerce, AI + healthcare, or AI + education?"

That sounds advanced, but the question is whether you genuinely understand the X after the plus sign.

He favors the reverse path:

Begin in a field you have worked in deeply for years, abstract your understanding into rules and processes, and then give them to AI for execution.

If you are a traditional Chinese medicine practitioner specializing in gynecology, you can build an intelligent Agent for gynecological consultations;

If you have spent more than ten years taking businesses into Arab markets, you can build a vertical Agent for Middle Eastern e-commerce operations.

Such an Agent need not serve millions of people. Serving several thousand high-value users with urgent needs and strong retention can already create considerable value.

He even offered a relaxed way to set goals: if you previously had to work 10 hours a day to earn 100,000 a month,

first try using AI to free yourself—work only 2 hours a day while maintaining monthly income of 100,000,

and spend the remaining time living and learning rather than immediately changing your goal to earning 1 million a month.

AI is not here to force you into more intense competition. It can instead help you compete more intelligently.

4\. Do Not Forget One Piece of Advice That Sounds Uncommercial but Matters

Amid the discussion of efficiency and conversion, Liu Rushan offered a view that sounded gentle but was in fact incisive:

AI is rapidly taking over every kind of dirty and exhausting task.

Two or three years from now, when machines perform all that repetitive labor, how will people make a living?

Her advice to creators and individual entrepreneurs is to begin becoming gentler, kinder, and more considerate now.

That may sound like inspirational rhetoric, but it is highly practical in business.

When productivity becomes sufficiently powerful, supply will inevitably exceed demand and demand will become increasingly specialized.

Users will no longer buy only functionality. They will buy an aesthetic, a set of values, and the feeling that this person genuinely cares about me.

You can possess a personal AI production studio, an automated product-selection system, and a complete Agent workflow.

What is genuinely difficult to replicate is your sensitivity to and judgment about the world:

What problems do you see?

Whom do you choose to serve?

What kind of story do you want to tell through these tools?

V. In an Era of Equal Access to Technology, Where Does the Moat Actually Lie?

When everyone can use roughly the same models and tools, the word "moat" may seem increasingly absurd.

Yet the guests shared a remarkably consistent consensus: the moat has never resided in flashy features.

Wei Lihua's perspective is simple: in several years, AI capabilities will be as widespread as Excel tools.

The factors that truly create separation will remain your understanding of users, refinement of product details, creative boundaries, and the small measure of soul you give a product.

Wang Qing divided the moat into two more concrete components:

The first is industry Know-how.

You must become middleware between AI and the customer, knowing which parts of the business AI can handle, which require human decisions, and how to combine the two into a smooth operational chain.

The second is accumulated data.

Can you structurally accumulate the data, feedback, and conversion outcomes generated by every interaction between AI and users, turning them into tacit knowledge that belongs only to you? If you merely connect a general-purpose model and add an attractive shell, you may disappear the day someone produces a cheaper version.

Shu Junliang added that you must first understand whom you are defending against.

Defending against giants such as Microsoft and Google is almost unrealistic. Individual entrepreneurs generally need only drive efficiency and cost to the limit in a sufficiently vertical niche to pull ahead.

Compressing work that once required 5 people into one person plus a pipeline is itself an advantage that peers will struggle to replicate.

Peng Yunye added a direction worthy of attention from Chinese entrepreneurs: AI + hardware.

He cited a recording-card product that addressed the difficulty of convenient recording on the iPhone, then added AI-generated meeting notes. A small team could build a business operating at a scale of more than US$100 million a year.

Shenzhen's many smart-hardware manufacturers are also integrating AI into products such as handheld gimbals, following similar logic: embody AI capabilities through hardware in the physical world.

Liu Rushan reminded everyone to beware of being emotionally deceived by a Demo.

Promotional videos for many AI products show a 10-point result while the real experience deserves only 1 point, because entrepreneurs are also telling stories to raise financing.

Rather than rejecting AI completely after one disappointment, try it again several months later. The cost of missing a technological path that can genuinely transform efficiency is much greater than falling into one Demo's trap.

She added an observation: today's global AI war is, to a considerable extent, competition between Chinese people in the United States and Chinese people in China.

Chinese entrepreneurs already stand in the world's first tier. We have every reason to be more confident and a responsibility to be friendlier toward one another.

VI. For Those Still Watching: Five Sentences, Five Paths

At the end of the roundtable, Wang Chaochao asked each guest to offer one sentence of advice to individual entrepreneurs hoping to use AI to expand overseas.

Together, the statements resemble signposts pointing toward the future:

Read fewer articles and do more. Do not treat AI as a stream of news to watch for entertainment. First build a workflow, create an Agent, and run an advertising campaign; then discuss whether it is worthwhile.

Do not confuse a Demo with an application; let outcomes speak. Flashy demonstrations can be intoxicating, but cash flow recognizes only conversion rates and profit.

Find the right positioning, move quickly in small steps, and validate immediately. After choosing the correct broad direction, do not hide away building a perfect product; test a minimum viable version first.

Go global; overseas markets are less intensely competitive. Many industries have reached the limit of domestic competition, while overseas markets remain in an early period of opportunity. The condition is that you must be willing to take the step.

Be a little gentler and kinder than before. When everyone can use roughly the same AI, those who remain will often be the people willing to listen carefully, serve sincerely, and communicate thoughtfully.

If you have read this far but are still debating whether to wait and observe, close this article, choose any small task, and use AI to transform it completely:

Ask AI to rewrite an English prospecting email;

Or build a simple video-generation workflow;

Or use an Agent to analyze your store's recent data.

When you clearly feel for the first time that you can complete the task alone and do it so much faster,

you have truly entered this wave.

As for what comes next—whether you become a super individual equipped by AI or build a small but beautiful global business with a few like-minded partners—there is probably only one answer: start moving before discussing strategy.

More Details from the Conversation

Guest Introductions

Wang Chaochao: Thank you very much to our five guests for joining the roundtable. Without further discussion, let us proceed in order and ask everyone to introduce themselves and their companies.

Wei Lihua: Hello, everyone. I am Wei Lihua, founder of HelloTalk. HelloTalk is a global cross-border language-social product with more than 60 million registered users worldwide, and the company is based in Shenzhen.

Wang Qing: Hello, everyone. I am Wang Qing, CEO of Boolean Vector. We primarily build an Agent for cross-border and global e-commerce users that can generate, produce, and edit advertising creative assets at scale.

Shu Junliang: Hello, everyone. I am Shu Junliang, a serial entrepreneur and currently CTO of MuleRun. MuleRun is an exceptionally new platform that launched only a little more than two months ago. Put simply, it is a "Taobao for AI Agents" or an "App Store for AI Agents." Our mission is to serve developers and users, help more useful Agents that solve real problems emerge around the world, and bring those Agents to more users.

Peng Yunye: Hello, everyone. I am Peng Yunye, a partner at QuickCEP. We built a system that provides overseas Agent customer service for cross-border e-commerce. When brands such as Pop Mart and Honor sell worldwide, their customer service on independent websites and marketplace platforms uses our tools. Thank you.

Liu Rushan: Hello, everyone. I am Liu Rushan, founder and CEO of Wilderness Stars. Our product Mulan.pro is currently the world's only product capable of "canvas-based workflow editing," and the only product that can build video workflows through natural-language commands.

Let me add two sentences of explanation. The first capability single-handedly moves AI video production from "handicraft" into "industrial assembly-line production." The second allows even a ten-year-old child to build a commercial-grade video workflow with natural language.

How Does AI Change the Threshold and Rules for Individuals Going Global?

Wang Chaochao: Our five guests represent different tracks: two work on content, one on marketing customer service, one on an Agent marketplace, and one on a voice-social platform. It is a wonderful combination. Today's theme is "e-commerce going global and the super individual."

Please use a customer or user case from your fields to explain which fundamental elements of the latest AI wave have raised or lowered the threshold for individual entrepreneurs to expand overseas, or even changed the rules of competition. Here, a "super individual" may mean a lone operator, a small team, or even a super-sized medium or large team.

Wei Lihua: I have seen many reports showing that companies with only one or two people can now build highly global products.

The greatest change brought by AI begins with language. In the past, it was difficult for Chinese individuals or small companies to enter overseas markets because they had to generate webpages, content, copy, and email marketing—overseas markets depend heavily on email marketing—and localization into languages such as Japanese and Korean was extremely difficult.

Today, however, if you can understand English roughly, you can use AI to translate all the content into different national languages very naturally. Even HelloTalk itself, with its enormous user base, now uses AI directly for many languages without requiring human proofreading. The change is considerable and has already become a core scenario.

Wang Qing: We serve two kinds of merchants: large customers and small customers. Small customers mainly resemble overseas retailers selling handicrafts in spaces similar to China's creative parks.

China's current environment is intensely competitive, and many friends ask whether they can expand overseas. We observe that AI has substantially reduced the cost of international e-commerce. Previously, entrants first failed the language barrier, then lacked editing ability, and third lacked filming skills.

Today, they can upload material filmed in China to a platform or use a tool to generate with one click various videos capable of producing orders, running as advertisements, or being posted to social media. Set aside whether the quality is world-class; the entry threshold has become extremely low. As noted earlier, a super individual can now easily go from nothing to something.

Wang Chaochao: Output undergoes a dramatic leap.

Wang Qing: Yes, exactly.

Shu Junliang: This topic closely matches our product positioning. As an Agent platform, MuleRun naturally supports small teams and super individuals.

Tasks that once required a team to expand overseas—creating product images, hiring models, filming video, and communicating across languages when listing a new product—can now be handled in a dedicated e-commerce area on our platform that brings together more than ten different Agents, covering every capability from image creation to data analysis.

Another example is that AI broadens the concept of "going global." Previously, it meant selling Chinese goods abroad. Today, an Agent itself is also a kind of "merchandise."

Here is a real example. On the day Google recently released an image-generation model—source editor's note: perhaps Imagen 3 or a similar model update—an independent developer on our platform sensed perceptively that it would become popular and immediately built an Agent. The Agent was extremely simple. Its Prompt may have contained only 20 characters: input a sketch and generate a rendered image.

He may have spent only one hour building it and listed it on the day the model launched. As an individual, he used that wave of attention to bring the Agent to global users through our platform, which handled payments and related matters. The Agent earned more than US$1,000 in one week. Although "riding a trend" may produce only short-term traffic, this is a highly representative example today of monetizing traffic rapidly.

Wang Chaochao: Soft products, Agents, can be sold, while hard products, physical goods, can also be sold rapidly through AI.

Peng Yunye: The central pain points of cross-border e-commerce are language and time differences.

The customers we serve are relatively large, and cross-border e-commerce is evolving into "overseas consumer operations." We also see super individuals and small teams using AI for presales shopping guidance.

The data shows that the higher the average order value and the greater the product complexity, the higher the conversion rate when AI serves users. This is human nature: when buying something expensive, users want comprehensive information about specifications and after-sales service. In the past, we were asleep when foreigners visited stores because of the time difference. AI customer service and shopping guidance now provide real-time responses, increasing user trust and naturally improving conversion.

Liu Rushan: Our product remains in public beta, and a large proportion of users work in e-commerce on TikTok.

Going global has "three main tactics," one of which is influencer marketing. Previously, a furniture merchant seeking influencer marketing first had to conduct BD to find an influencer, send samples to the United States, communicate about filming and editing, and overcome time differences. The process generally took six months.

None of this is necessary now. You generate a video yourself with AI—our product offers high controllability—then ask the influencer, "May I swap your face into it?" The other party says OK; you send the face-swapped video, and they publish it on their own account. A matter that once required six months of communication can now be resolved in half a day.

There is another approach for a super individual. The Canton Fair recently ended, and everyone remarked that times had changed. In the past, a booth was enough for foreign trade. Today, overseas Buyers are highly professional: they check whether you have an independent website and care even more about whether you continually update videos on overseas media such as TikTok.

For a team of two or three, continuously producing video is expensive. With our workflow product, however, a seller sending fitness equipment to Brazil or Vietnam need only build a workflow and click generate. Every resulting video has the same music, lines, captions, and advertising-copy framework, but the people and actions differ each time. This is critical because platforms use duplicate-detection mechanisms and do not permit montages of repeated content.

We call this "familiar yet different," or "similar but not the same." The initial setup may take half a day; subsequent videos require only ten minutes, and even a ten-year-old child can learn the process and keep producing. This is disruptive industrial production.

Wang Chaochao: This has practically become a large-scale live sales demonstration.

Deployment in Practice: Which Core Areas Merit AI Investment, and What Are the Pitfalls?

Wang Chaochao: Individuals have limited resources, including time, money, and energy. In your view, in which core areas can AI investment achieve disproportionate leverage? What pitfalls may arise while embracing AI?

Wei Lihua: There are two core areas:

Market research and product validation: AI makes it easy to ask whether similar products exist and understand competitors.

Marketing and promotion: applications range from creating copy and creative assets to using AI to scrape potential users on platforms such as Instagram—for example, a nail-art business scraping the followers of nail-art influencers. This previously required programming but can now be automated with AI. There are certainly pitfalls, mainly involving details, but with new models such as Gemini being released, the overall problems should be manageable.

Wang Chaochao: In summary, these are the two ends of the smile curve: insight and product definition, and sales and promotion.

Wang Qing: My view is somewhat different. AI now claims to solve every problem, but doing so is actually very difficult.

We work in a specialized field and must focus on outcomes. In advertising placement, for example, customers care most about ROAS (return on ad spend).

Critical area: the first 3 seconds of a video are the most important. If AI selects unusable footage while screening material, it wastes money. Advertisers care deeply about this issue, and AI does not yet handle it perfectly.

Greatest pitfall: confusing a foundation-model Demo with an application, or Product. A Demo often looks only 1% away from perfection, but that 1% may either be impossible to cross or actually require another 70% of the effort needed for commercialization. Entrepreneurs who mistake a Demo for an outcome will suffer greatly.

Shu Junliang: I believe we should not think in terms of "AI + X," but reverse the order and think in terms of "X + AI."

Do not search for nails while holding a hammer. Build from the field you know best as a vertical expert and use AI to solve problems in that field.

Our platform does not recommend generic AI Coding or PPT tools, which major companies have already validated. We hope instead to see vertical Agents. If you are a traditional Chinese medicine practitioner specializing in gynecology, or an expert in taking businesses into Arab markets, you can turn that Know-how into an Agent. It need not serve millions of users; finding several thousand core users creates considerable value.

My suggestion is this: if you previously worked 10 hours a day to earn 100,000 a month, use AI so that you can work 2 hours a day, continue earning 100,000, and spend the remaining time living. Do not focus on earning 1 million a month; people should relax more.

Wang Chaochao: In summary, the nail still exists, but I used AI to build a different hammer. Work backward from needs and outcomes.

Peng Yunye: I strongly agree with Mr. Wei's view: use AI to understand overseas demand.

For example, one of our customers lived in the United States and discovered that getting eyeglasses there involved a long process and required an appointment. He came from Danyang, China's eyeglass manufacturing base, and combined his resources with foreigners' pain point to build a cross-border eyeglass e-commerce business that now generates 2 billion a year (currency unspecified in the source).

Another customer discovered that people in some states cultivated special plants and wanted to grow them at home, so he built a cultivator.

These demand insights are exceptionally important. Companies in the industry such as SellerSprite are also combining AI with product-selection analysis for merchants. Without overseas living experience, you must use AI to develop insights and identify categories that combine with your resource advantages.

Liu Rushan: As a female entrepreneur, I may take the discussion a little further.

People previously engaged in short-term behavior to survive. Now AI has improved productivity. If AI can perform the dirty and exhausting tasks, what will you do two or three years from now?

My advice is to begin becoming a gentle, kind, and considerate person. That is your uniqueness.

Previously, demand focused on mass-market goods. As productivity rises, demand will become highly specialized. The key to future success will be personal character, aesthetics, and inner kindness and consideration.

First quiet yourself and show consideration for others; only then can you discover specialized needs. Stop competing on low quality. Once you possess a "personal AI production studio," what matters is what you want to express through it.

Wang Chaochao: The preceding four "straight-talking men" were discussing nails and hammers, and Ms. Rushan suddenly made the roundtable very gentle. Yet all paths lead to the same destination: starting from needs.

The Moat: How Can Individuals Build Core Competitiveness That Cannot Be Replicated?

Wang Chaochao: AI accelerates equal access to technology and tools are converging. Will competition become homogeneous? How can individual entrepreneurs use AI to build a moat?

Wei Lihua: After some time, AI will become as widespread as computer skills.

The real moat will remain user insight, the details of product refinement, creativity, and the product's "soul."

AI merely makes capable people more efficient. Whether a product truly wins the market will still depend on the storytelling ability and product understanding of the person behind it. This is not greatly different from the past.

Wang Qing: I agree. The more you understand AI technology, the clearer it becomes that general-purpose and flashy things are not moats.

There are two genuine moats:

Industry Know-how: your understanding of the business, customers, and industry. You are the bridge, or middleware, between AI and customers. You must know which parts AI can perform, which it cannot, and how to maintain control.

Accumulated data: can you preserve the data and user feedback generated after AI runs and create a proprietary closed loop? If it is only a tool, you disappear when someone releases a cheaper alternative.

Shu Junliang: First define whom the moat protects you from. Defending against Microsoft and Google is unrealistic.

For individuals, rather than expecting AI to create something entirely new, use AI to make your existing processes simpler, more automated, and less expensive.

Turning steps such as video tagging and article translation into automated pipelines creates an efficiency and cost advantage that functions as a moat in the medium term. Over the long term, AI is developing so rapidly that perhaps AGI will arrive one day and everyone will lose their jobs, but that is another matter.

Peng Yunye: I think AI + hardware may become a threshold for individual entrepreneurship in the future.

For example, a product called Plaud appeared three years ago. It is an AI recording card that solved the iPhone's inability to record conveniently and could also organize meeting notes. By combining AI with hardware, even a small team could generate US$100 million a year.

Hohem in Shenzhen, which makes handheld gimbals, is also integrating AI. Shenzhen offers many such opportunities: combining AI with hardware and using a first-mover advantage to create a breakout market may constitute a barrier.

Liu Rushan: Let me add one pitfall. Promotional Demos for many AI products show a 10-point experience while the reality is only 1 point, leaving users disappointed.

Everyone should understand that entrepreneurs may exaggerate somewhat to secure financing. Do not give up because the first use disappoints you. Try again after three months or half a year; it may work then. The cost of missing out is very high.

I spent a long time in Silicon Valley and found that today's AI war is really a war between "Chinese people in the United States" and "Chinese people in China." We need national confidence. Chinese AI entrepreneurs are already world-class, so please be nice to us.

One-Sentence Advice

Wang Chaochao: Finally, please give individual entrepreneurs one sentence of advice.

Liu Rushan: Be kinder and gentler than before.

Peng Yunye: Go global; overseas markets are less intensely competitive.

Shu Junliang: Find the right position, move quickly in small steps, and validate immediately.

Wang Qing: Do not confuse a Demo with an application; let results speak.

Wei Lihua: Read fewer articles, do more, and get moving.

---

Original publication: https://uniqueresearch.substack.com/p/src-20251205-01html
On-site reading page: https://ffcap.cn/en/research/src-20251205-01html