---
title: "15 People, a US$100 Million Target: The Boss Says No Hiring for Three to Five Years"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2026-04-11T12:01:22+00:00"
canonical: "https://ffcap.cn/en/research/src-20260411-01html"
source: "https://uniqueresearch.substack.com/p/src-20260411-01html"
language: "en"
---

# 15 People, a US$100 Million Target: The Boss Says No Hiring for Three to Five Years

_Original · Unique Research · 2026-04-11_

_Editor's note: The US$100 million figure is an unnamed customer's target, not an independently verified achievement; the English headline makes that distinction explicit. This translation preserves the source author's first-person narrative and the panelists' accounts, not the translator's firsthand experience. Headcount descriptions differ across passages, the narrative calls Ning Liaoyuan a founder while the roster calls him CTO, and the narrative and Wang Baochen's transcript differ on whether export volume replaced earlier metrics or was always central. These differences remain visible pending verification. Relative dates refer to the original April 2026 discussion. Company metrics, legal-service accuracy and delivery claims, fees and enterprise counts are source reports, not independent audits or legal advice. Statements about Transformer limits, job displacement and future earnings are attributed views, not established guarantees. Source images and final publication checks remain outstanding._

Unique Awards

After Listening to This Hangzhou AI Roundtable, Layoffs Were Not What Frightened Me Most

What is more unsettling is not people being replaced, but entire industries becoming willing to pay only for “done.”

"

Let me tell you a horror story.

A company with 15 people has a target of one hundred million US dollars.

After hiring the final two or three people, the boss says the company will not need to recruit again for three to five years.

That is what TTC founder Ning Liaoyuan said at the Hangzhou AI WEEK roundtable.

One of his clients, a company with 15 people, had a target of one hundred million US dollars. After hiring the final two or three people, the boss told him there would be no need to recruit for three to five years.

Sitting in the audience, my first reaction was: That's impressive. My second was: Damn, where do all the other people go?

After listening to the whole session, I realized that horror story was only the beginning.

Wu Bin of Jirui Technology said his company had cut its workforce from 180 to 60. Meanwhile, Yu Chunyan of Xuntu Technology said her company had grown from 90 people last year to 123 now, with plans to reach 150 this year.

One side was cutting people. The other was competing to hire them.

That completely threw me.

Frankly, I had assumed the AI era meant straightforward substitution: Higher efficiency, fewer people. But after listening to the five guests, I realized things were not that simple.

Some Cut Staff, Others Expand. The Difference Is Not AI, but How You Use It to Make Money.

Start with Wu Bin. He provides AI services for e-commerce. The company had 180 people the year before last, more than 90 last year, and now just over 60.

How did he reduce the headcount?

AI Coding enabled cuts in product and engineering. AI capabilities reduced a delivery team that had numbered thirty or forty. Sales went from 80 people the year before last to 17 now.

He now builds his own personal brand, or IP, with 1 million followers. AI helps write scripts and make videos. He said meeting payroll every month had made him anxious the year before last. Now, with 60 people and a much smaller payroll, more becomes profit, and he feels more secure.

I asked whether he even needed to keep those 17 salespeople. He did not deny the possibility.

But Yu Chunyan's story went in the opposite direction.

Xuntu's product, Alpha Pai, serves institutional investors. More than 7,000 institutions and over 80,000 people worldwide use it. The company had fewer than 100 people last year and plans to keep growing to 150 this year.

“We have been growing all along.”

I asked whether that meant lower productivity per employee.

She said they have only twenty or thirty salespeople. A traditional competitor serving that many users would need at least a hundred. Their sales productivity is 5–10 times the traditional level.

Let me tell you where the difference lies.

Chuangkit founder Wang Baochen offered a clue.

They build AI design tools. Their north-star metrics used to be DAU and MAU. Now they look at the number of content exports.

In the tool era, an individual exported a few dozen pieces per week. In the AI-agent era, the platform generates nearly 10 exports every second.

That change in metrics says a great deal.

They used to care how long you used the product. Now they care what you completed. AI has eliminated “time spent looking busy” as a metric.

But what followed was even more interesting.

“AI is an amplifier. But what actually gets amplified is your distinctive assets and consistent voice, provided they can be reused.”

It may sound harsh, but it is true.

For a mediocre designer, AI makes the work more mediocre. For a designer with a distinctive style, AI can make them extraordinary.

The Bigger Shift Is Not Layoffs, but a New Unit of Pricing for Professional Services

NomiLaw founder Chen Ming gave an example.

A company with assets and liabilities both in the hundreds of millions was on the verge of insolvency and needed a comprehensive legal and tax diagnostic report. Traditional delivery would take at least three or four days, with legal and tax teams working separately. Using his product, he handled input, output, verification, and proofreading alone in less than two hours. He checked each cited rule and reference case: “Not one piece of substantive content needed changing.”

I asked about the difference in fees. He did not give a specific number, but the implication was clear. Professional services used to be priced in “hours.” What becomes the unit in the AI era? Total fees often used to run into “tens of thousands.” What happens to that in the AI era?

But Ning Liaoyuan offered a caution.

“The limits of the Transformer architecture are very clear. Another 1.5-fold to 2-fold improvement is probably about as far as it goes. The next technological revolution will require waiting a few more years.”

He was not dismissing AI. He was saying it has a ceiling.

That ceiling is precisely why people can become valuable again.

The People Whose Value Rises Are Those Who Know How to “Ride the Horse”

His company, TTC, recruits AI talent. Early in the year, clients were telling horror stories: Productivity had surged, and they would not recruit for three to five years. After the Spring Festival, everything changed. Companies that had said they needed no more people began saying they needed many.

“When AI takes your productivity per employee from 500,000 to 3 million or 5 million, market demand grows even faster. The boss next door is trying to take your business. You have to hire more people.”

But they are not hiring the same kinds of people as before.

They need people who know how to ride the horse—people who can direct AI and lead AI teams.

Those people's incomes are shifting from a normal distribution to a power-law distribution. AI compresses the middle and lifts the top.

I suspect the idea that there is no middle ground will become increasingly real this year.

As the roundtable approached its end, I asked each guest the same question: What kind of person will not be displaced in the future?

Ning Liaoyuan offered one word: Passion.

“AI's capabilities are limited. People whose understanding goes beyond AI have their abilities amplified further. They no longer need to write weekly reports or organize documents. When you have distinctive talent and understanding, your income rises substantially. Fundamentally, it comes down to loving what you do. Without Passion, you do not get results.”

Yu Chunyan talked about tolerance for mistakes.

“Finance is the industry most readily transformed by AI. It will replace 80% of desk work. But what does not change is a person's level of understanding. If yours is not high enough, you cannot train a good AI assistant. Evolution's ultimate goal is not perfection, but continuing to survive.”

Wu Bin emphasized generalists.

“A single specialized skill can easily be overturned by a model update. Internally, we think the people who survive will be programmers who understand sales and salespeople who understand code. Know a little about each skill, and with AI you can understand a great deal.”

Chen Ming offered a four-step loop.

Define the problem, verify the facts, exercise professional judgment, and choose values. Those are four things AI can never replace people in.

Wang Baochen summarized it as “three plus one”: One vertical line—understanding AI—runs through everything. The three horizontal lines are good health, imagination, and logic.

No one mentioned working hard. No one mentioned overtime.

Thinking it over, every answer pointed in the same direction. The scarcest things in the AI era are not skills, but human initiative and judgment.

What Is Most Frightening Is Not Layoffs, but the Declining Value of “the Process”

Return to Ning Liaoyuan's statement: In the future, there will be no human workhorses, because workhorses do not create value.

But look at Wu Bin's situation.

He had livestreamed until 12 the previous night. Viewers kept posting: “Brother Bing, end the stream. You're working too hard.” He said it was tiring but worthwhile, because it saved a lot of money.

In the AI era, the boss becomes the workhorse personally, then eliminates the need for the former workhorses.

The workhorses are gone, but the boss is now turning the mill.

Is that liberation or a trap?

Honestly, I am not sure.

All I know is that the scariest part of the “15 people doing one hundred million” horror story is not the 15. It is the one hundred million.

Finally, I asked Wu Bin what his goal was in running the company.

He said that in the short term, at least, he wanted to take it public and deliver for investors.

I did not respond. The audience laughed. Then everyone fell silent.

In that silence was something everyone understood but nobody said.

Once AI pushes efficiency to the limit, “delivering for investors” can become another horror story. The boss uses AI to get 60 people to do the work of 180. Profits rise—but where do the 120 people outside that group of 60 go?

The other side of efficiency is substitution.

When Wu Bin said, “Service providers have spoiled them; there's nothing we can do,” he meant merchants increasingly want results and will not pay for the process.

Change the subject of that sentence and it still works: AI has spoiled them; there's nothing we can do.

When AI can produce a 78-page legal and tax report in two hours or generate 10 designs every second, who still wants to pay for “I'm working on it”? They only want to pay for “It's done” or “It's sold.”

Process is losing value. Results are becoming more expensive.

That trend is more frightening than layoffs. Layoffs are transitional pain; a change in pricing is an earthquake for an industry.

More from the Conversation

Unique Awards · Hangzhou AI WEEK Trend Roundtable Panel

“Learning and Knowledge: AI as a ‘Second Brain’ for Knowledge Management and Greater Efficiency”

Guests:

Wang Baochen, Founder & CEO of Chuangkit

Chen Ming, Founder of NomiLaw Nuomibao

Yu Chunyan, Co-founder of Xuntu Technology

Wu Bin, CEO of Jirui Technology

Ning Liaoyuan, CTO of TTC

Moderator: Wu Wei, Founder & CEO of Unique Research

Wu Wei: First, let me explain my moderating style. I have not prepared any questions for you in advance, so all my questions are of the moment. I think that is part of why people came: We want an immediate response. If questions are set beforehand, answers can be generated beforehand too. So we are generating them in real time, using each person's “second brain,” to give the audience the most practical insights. Some people may not know you very well, so let's introduce each company in the simplest possible sentence. I also have a key question embedded in your introductions: What is your company's north-star metric now, and why? Introduce the company, then answer that.

Wang Baochen: Hello, everyone. I'm Wang Baochen from Chuangkit. We have consistently built AI design tools. People working in corporate marketing and new media may have used them. Through AI, an intelligent editor, and a large library of licensed resources, we create visual content such as designs and videos. That is what we do today. Chuangkit is also focusing on developing and exploring AI agents for visual creativity.

Wu Wei: What is your company's north-star metric?

Wang Baochen: We have always been a productivity tool, so our north-star metric has consistently centered on customer exports—the volume of content exported.

Wu Wei: The number of content exports, right.

Wang Baochen: Whether content was previously generated with tools or is now generated using AI and agents, we care more about whether users actually use it: export it, share it, and put it to use.

Wu Wei: If your metric is exports, does that mean the work has been completed within your system?

Wang Baochen: It means that stage of value has been created. Of course, we are also building subsequent extensions, but this part demonstrates that the user is paying for that value.

Wu Wei: I think that is a good point: Your core metric is total output. Roughly what scale are we talking about?

Wang Baochen: Previously, when it was purely a tool, the average individual user could export several dozen pieces each week. Now, with AI, we see nearly 10 exports per second.

Wu Wei: Per second, right—ten per second across the entire platform. An average user can export about 10 per week. Why does it feel like less than I would have imagined?

Wang Baochen: If I am a designer, I might reach ten in a day. I had not finished explaining: With AI agents, that figure is rising quickly. Previously, users worked with the tool; now they use AI and then export.

Wu Wei: OK, that is useful to think about. It may be central to the next generation of software. You no longer consider how long users stay in your software; you may care more about how many results they complete. OK, Mr. Chen, introduce yourself and answer the question.

Chen Ming: My name is Chen Ming. I have worked in law for 20 years and am the founder of NomiLaw Nuomibao. NomiLaw aims to combine AI's information-service capabilities with experts' professional human services for compliance, creating an AI platform with human–machine collaboration and clear boundaries of responsibility.

Wu Wei: Could it work without lawyers?

Chen Ming: Without lawyers, it has only pure AI capabilities, which objectively cannot guarantee 100% accuracy. Although NomiLaw can currently achieve 98% or 99% accuracy, this is a highly professional, specialized, and serious setting that requires 100%, so we provide human support.

Wu Wei: So a Human professional still needs to be part of the task-interaction loop, perhaps performing something like a confirmation or rejection.

Chen Ming: Yes. That is very similar to the healthcare topic we discussed above.

Wu Bin: I'm Wu Bin from Jirui Technology. We are a service provider focused on e-commerce and have always helped merchants create content. Recently, we have seen AI become stronger and stronger, making a software-only service feel too thin. So we also moved into content delivery, with our team directly producing TVC advertisements, commercials, and sales videos for merchants. Starting last year and the year before, we also began helping merchants sell products directly, providing a complete service. Merchants increasingly focus on Revenue, so we want to provide end-to-end sales value, using AI content to grow e-commerce revenue. We are a marketing-driven company. Our internal north-star metrics have recently received a lot of attention: the number of merchants served per employee and the GMV growth we generate for them.

Wu Wei: Output per person.

Wu Bin: Yes, the GMV growth each person generates for clients. We increasingly value that metric because our own headcount keeps falling, and we use a great deal of AI to improve efficiency. Last year and the year before, we focused on substitution rates: How many people did content creation need, and how many videos could one person make for a merchant? Recently, we have found that what merchants really want is the final GMV. That is also part of our assessment.

Wu Wei: Aren't merchants asking too much? They always want the final result and refuse to pay for the process. Why?

Wu Bin: I think there is nothing we can do. Service providers have spoiled them.

Wu Wei: So you spoiled them. What is that metric roughly now?

Wu Bin: It is still rising. For a strong client, we might generate an additional thirty or forty million, or fifty or sixty million, in annual GMV, perhaps through just one or two people.

Wu Wei: Incremental GMV?

Wu Bin: Yes, incremental.

Wu Wei: OK, that looks good. How many people are in the company now?

Wu Bin: We have come down to just over 60 this year.

Wu Wei: How many last year?

Wu Bin: More than 100 last year.

Wu Wei: Then let's discuss layoffs shortly.

Yu Chunyan: I'm Yu Chunyan, co-founder of Xuntu Technology. We have a product called Alpha Pai, mainly for institutional investors. It provides AI-driven information, data, and productivity. Investment managers, researchers, and professional institutional investors who buy financial products previously relied more on specialist data terminals. We position ourselves as their AI investment-research assistant. We launched the product after AI emerged in ’23. We now serve roughly 7,000 institutions and more than 80,000 institutional investors worldwide. In AI for finance, we rank among the leaders globally in user engagement.

Wu Wei: I think that is an excellent starting point. Early in AI's development, meeting transcription was one of the best use cases, and your audience can pay. What were you thinking at the time?

Yu Chunyan: Our partners already worked in finance and came from data and AI teams. There were no large models then, so we relied more on traditional technologies. We have always followed technological iteration. Whenever a technology reaches a breakthrough point, your accumulated resources, Know-How, and data can come together at that moment. From Day 1—from ChatGPT appearing in ’23 to OpenClaw becoming a major topic over the past two months—we have applied technology to our research itself. Our Slogan is “One for everyone in Lujiazui.”

Wu Wei: One for everyone in Lujiazui—so if someone there still isn't using Alpha Pai, they are basically outside the mainstream circle.

Yu Chunyan: On the north-star metric you mentioned, it has changed every year over these three years. In the first year, we looked more at DAU and MAU, closer to traditional internet metrics. Last year, as we added more AI Agent applications, we focused more on how many active users used AI capabilities and features every day.

Wu Wei: How many AI functions and capabilities your users use.

Yu Chunyan: Initially, people mainly came to consume content, such as meetings and research reports. But after we introduced more Agent features last year, we focused more on how many daily and monthly active users were frequently using our AI capabilities.

Ning Liaoyuan: My name is Ning Liaoyuan, and my company is TTC. TTC recruits AI talent. Our slogan is “For AI talent, come to TTC.” Whether you are looking for a co-founder, a company, or outstanding talent, you can come to us. Our north-star metric has also changed. At the very beginning, it was the “number of high-quality résumés.” Last year, we essentially focused on the “number of first interviews.”

Wu Wei: The number of initial interviews. No matter how many people you recommend to a client, at least when they accept one, they want to have a conversation.

Ning Liaoyuan: Their willingness to talk is a good beginning. A person and a good opportunity start a conversation, and possibilities follow. I actually considered charging clients directly by the number of first interviews.

Wu Wei: I think what matters most to clients is value—results. Making the client bear the uncertainty themselves does not seem very good.

Ning Liaoyuan: In fact, many recruitment platforms have recently begun charging for results. Charging for a process that produces no result is nonsense.

Wu Wei: Will recruitment websites still exist in the future? Will we still need them?

Ning Liaoyuan: I think we may still need them, but they will be less useful.

Wu Wei: Why?

Ning Liaoyuan: Connections and information between people cannot be delivered through that traditional recruitment-website model alone.

Wu Wei: How will you change that model?

Ning Liaoyuan: That may be the next episode. In the Agent era, recruitment itself may change, not just recruitment websites. People's talent may no longer be expressed by physically joining a company and tolerating colleagues they do not particularly like. People may connect directly through other channels, such as an Agent or other service formats. Many OpenClaw setups are already, in a sense, working for multiple companies at once. A person's talent can serve several companies through a new Agent or service format, without that person having to interview and join each company.

Wu Wei: So you mean an Agent will work for me or find me a job, and I simply need to use it well.

Ning Liaoyuan: I see an Agent more as a vehicle and amplifier for human talent. It is not that it works for you; you work through it, except that you can sleep while doing so. Its talent or capability comes from what you give it. You must keep improving it, because if you move more slowly, someone else will overtake you.

Wu Wei: So we are the workhorses now, and AI will be the workhorse later.

Ning Liaoyuan: Based on my current thinking, there should be no workhorses in the future, because workhorses do not create value. If someone earns money by working like a workhorse, they probably will not have opportunities in the future. If someone works from their interests and distinctive understanding, the opportunities will be enormous.

Wu Wei: I don't think being a workhorse is a good state either. We've had a little Warm up; the questions will get tougher now. Mr. Ning, how many people are in your company?

Ning Liaoyuan: We currently have 200 full-time employees.

Wu Wei: How many last year?

Ning Liaoyuan: A little over 180. We made some adjustments, laying off some people and hiring others.

Wu Wei: So you added some and reduced others. What kinds of people did you let go, and what kinds did you hire?

Ning Liaoyuan: Let me first describe a trend I've observed. In the second half of last year and early this year, many of our clients were making layoffs, some quite substantial.

Wu Wei: Then how do you do business? Your clients are laying people off, yet you help them recruit.

Ning Liaoyuan: One day, a particularly strong client with just over a dozen people suddenly asked us to hire two or three more. He said: “Liao, let me tell you a horror story. After hiring these last two or three people, I think we won't need to recruit for three to five years. Our 15 people will be enough.”

Wu Wei: Their target is one hundred million US dollars, but they think 15 people are enough.

Ning Liaoyuan: Quite possibly. In January, we thought that horror story really described what was happening. But after the Spring Festival in February, things changed. Many of those companies began saying they wanted to hire lots of people, including us. In January, I spoke with our CEO about how much efficiency had improved and whether we could cut some business staff because we would not need so many this year. Then things changed.

Wu Wei: What is the biggest difference between the people you want now and those you wanted before? What are the most characteristic traits or profiles you need?

Ning Liaoyuan: We need two kinds. One is people who can direct AI or AI teams—people who know how to “ride the horse.” We also need experts with sound judgment and industry experience. I can explain why many companies have started hiring again. Once AI provides additional capability, if productivity per employee goes from 500,000 or 800,000 to 2 million, 3 million, or even 5 million, you need to hire more and more people.

Wu Wei: Why not raise productivity per person even further? Can it not improve more?

Ning Liaoyuan: I've noticed that some recent speakers discussing OpenClaw have mythologized AI a little too much. The capability limits of the Transformer architecture are very clear. Another 1.5-fold to 2-fold improvement is probably about as far as it goes. The next technological revolution will require waiting a few more years.

Wu Wei: People can feel that the boundaries have become relatively clear.

Ning Liaoyuan: So I don't think productivity per person will grow indefinitely, but market demand is exploding. The boss next door is preparing to take your business, and that brings us back to competing for people.

Wu Wei: That may be a somewhat counter-consensus observation. Understood. At Xuntu, has staffing grown or shrunk over the past few months?

Yu Chunyan: We have been growing consistently. We had fewer than 90 people last year, are now at 123, and plan to reach 150.

Wu Wei: Both of them are adding people. Bin, how much has your headcount fallen?

Wu Bin: We had 180 the year before last, more than 90 last year, and just over 60 now.

Wu Wei: Look at that. Do you think this happened naturally?

Wu Bin: Partly naturally, and partly because AI became stronger. We previously had a delivery team of thirty or forty, which OpenClaw capabilities have now reduced. AI Coding also reduced product and engineering headcount. Then there is sales: We had 80 salespeople the year before last. I checked today, and we now have 17.

Wu Wei: I feel you don't even need 17. Just keep yourself and let AI sell.

Wu Bin: We still need people to meet, socialize, and communicate. I've begun building my own personal brand and now have nearly 1 million followers, which has brought many resources. I think it is easier for bosses to build personal brands now. Every morning, AI writes a script for me. I used to read it myself and send it to my director and editor. Now we are gradually trying to have AI make the videos too, and the results are quite good.

Wu Wei: I've noticed something about you: Many bosses may not be so Hands-on, but you are personally Hands-on. Do you think that distinguishes different founders?

Wu Bin: It takes time and is genuinely tiring. I livestreamed until 12 last night, and viewers commented: “Brother Bing, end the stream. You're working too hard.” But internally, we feel it is worthwhile because it saves a lot of money. Meeting payroll every month made me anxious the year before last. Now, with 60 people, payroll is much lower. When that becomes profit, I feel far more secure. It just makes me more tired personally.

Wu Wei: What is your goal in running the company?

Wu Bin: In the short term, at least, take it public and deliver for investors.

Wu Wei: Still about delivering for investors. Mr. Chen, I won't ask about recruitment, because what you are doing seems to let every individual and small company access AI legal services. How can AI perform a substantial part of professional legal work? Take me, for example. I retain a legal adviser every year for 100,000 yuan, but later realized most of the work was editing contracts. I felt I knew the provisions better, but did not dare change them. Once AI arrived, I gave agreements to a large model to revise. If they looked fine to me, I sent them to clients.

Chen Ming: It's fine; our audience is small, and not many people are watching. In a real case, we conducted a comprehensive legal and tax diagnosis and produced a written analysis for a company with assets and liabilities both in the hundreds of millions, on the verge of insolvency. Traditional work would normally take until at least the third or fourth day, with legal and tax work performed separately. With NomiLaw, I genuinely completed delivery alone in less than two hours. An analysis of roughly 78 pages was generated in tens of seconds, with cited provisions and reference cases. I checked each one afterward and did not change any of them.

Wu Wei: Doesn't that threaten the livelihoods of many lawyers and tax advisers?

Chen Ming: That topic is more suitable for an offline conversation, but the trend is irreversible.

Wu Wei: Undoubtedly. If I had lawyers and tax advisers work on it for a long time, roughly how much would it cost?

Chen Ming: Perhaps 100,000 or 80,000 would be normal. China used to have around 53 million micro, small, and medium-sized enterprises; now there may be around 60 million. This enormous group has genuine compliance, legal, and tax needs that were previously unmet. Human services could not cover it because of limited ability to pay and information asymmetry. Cutting prices to meet the demand was not realistic either. But with AI, combining it with human services can cover those needs.

Wu Wei: Making professional services broadly accessible. What is your most fundamental motivation for doing this?

Chen Ming: It begins with making compliance broadly accessible. The possibilities are substantial. It could become compliance infrastructure extending beyond law and tax: an AI workspace, a human–machine interaction system, or a system that enters the workflows of large companies or even government.

Wu Wei: You meet hundreds of entrepreneurs each year. Do you think Chinese entrepreneurs face more physical, psychological, or financial pressure?

Chen Ming: Physical pressure is a small Case. The main issue is psychological pressure, which of course relates to finances: so many employees, anxiety over their livelihoods, family concerns, and so on.

Wu Wei: Back to Mr. Wang. We discussed completed-task volume. There are many design tools, but people may not use these software products in the future. You also said AI may be the user. What is the best future form for design software?

Wang Baochen: Our original tools were mainly for people. Chuangkit changed something: Photoshop had a steep learning curve. In the independent-media and e-commerce era, many non-designers also needed to create content, and Chuangkit helped them. The core was making interaction easier through usable templates.

Wu Wei: But templates generated directly by AI may now be better than predefined ones.

Wang Baochen: Exactly. We think the future form will be designed not just for people but for an Agent. Enterprises have standards and specifications for commercial content. Previously, those became corporate assets, templates, and VI—visual identity. We believe these elements will be personalized to each enterprise, perhaps existing as an Agent's long-term memory.

Wu Wei: Does OpenClaw's arrival make you anxious?

Wang Baochen: Frankly, we see more positives than causes for anxiety. OpenClaw accelerates this shift. We think the ecosystem position for most enterprise-intelligence providers or startups is not to build the entry point, but the professional last mile: a specialized Agent. The little lobster—OpenClaw—has become an entry point, but today it cannot solve genuinely productive use cases by itself. Accumulated underlying capabilities and data are what make something sufficiently professional, rather than merely a visual model.

Wu Wei: This relates closely to AI as our second brain. Why choose this theme? Many creators, lawyers, and designers use AI as a second brain and store memories there. I no longer start anything from scratch; it feels as though AI has hollowed out my intelligence. As AI increasingly becomes our second brain, how should we, as individuals and professionals, relate to it? Mr. Ning, your thoughts?

Ning Liaoyuan: I've observed that this wave of AI still has limited capabilities. Because of those limits, people whose understanding goes beyond AI are having their abilities amplified more. I think even with more time, this wave of Transformer technology will not achieve certain cognitive breakthroughs; people will still be necessary. Human value, particularly the value of top people, is amplified. They no longer need to write weekly reports or organize documents. Where AI is particularly capable, practitioners' incomes are shifting from a normal distribution to a power-law distribution. Like influencers and stars, if you have distinctive talent, taste, understanding, and adaptability, your income rises substantially.

Wu Wei: Name the single most important trait you think the best future AI talent needs.

Ning Liaoyuan: Continuous improvement through learning and exploration. At the root is Passion. Without Passion, you will not get the result.

Wu Wei: Thank you, Mr. Ning. Let's give him a hand. His keyword is Passion.

Yu Chunyan: Finance runs on information, so we have always thought it the industry most readily transformed by AI. What changes is how you evaluate people, because perhaps 80% of desk work will be replaced by AI. What does not change is a person's cognitive ability. Particularly for an investment manager, if their understanding has not reached a sufficient level, training a good AI assistant is difficult. First, you still need firsthand information, such as speaking with entrepreneurs and interpreting their expressions. AI cannot replace that. Second, social actors are defined by various social relationships, which AI cannot replace either. We address the consensus layer, and users need to form their own non-consensus understanding on top of it. Finally, there is the ability to accept mistakes and keep experimenting. Models themselves are corrected through continual tolerance for error. Evolution's ultimate goal is not perfection, but continuing to survive.

Wu Wei: Survive. OK, a round of applause for Ms. Yu. The remaining three can address the second brain or talent traits.

Wu Bin: At this point, relying on a single specialized skill is very difficult. In the large-model era, a skill can be overturned as a model becomes stronger in one update. Internally, we think the people more likely to survive are generalists. They may not be particularly strong at each skill and know only a little, but with AI they can understand a great deal. We think programmers who understand sales, or salespeople who understand code, may be better roles in the future.

Wu Wei: If you had to define it in one keyword...

Wu Bin: General knowledge, or collaboration.

Chen Ming: We thought carefully about this when developing the product. First, defining the problem—AI cannot replace that. Second, verifying the facts. Third, professional judgment: Is the final AI result correct and consistent with reality? Fourth, value judgment: Ultimately, striking a balance based on the AI result. Defining the problem, verifying facts, exercising professional judgment, and making value judgments.

Wang Baochen: We describe human attributes as “three plus one,” or “three horizontal lines and one vertical.” The vertical line running through everything is understanding AI—you must know how to use it. The three horizontal lines are, first, good health and physical stamina as the foundation; second, imagination or inspiration; and third, logic or architectural thinking. Design is a non-standardized industry. AI as a second brain is indeed an amplifier, but the core of what gets amplified is your distinctive assets and consistent voice, provided they can be reused.

---

Original publication: https://uniqueresearch.substack.com/p/src-20260411-01html
On-site reading page: https://ffcap.cn/en/research/src-20260411-01html